The dog food label glossary

Every US dog food label carries a guaranteed analysis: minimum crude protein, minimum crude fat, maximum crude fiber and maximum moisture, expressed as percentages of the food as it comes out of the bag or can. Brands may add further guarantees, and must add one for any nutrient they make a claim about on the package.

Two things follow from the word guaranteed. First, these are limits, not measurements. A food guaranteeing 26% minimum protein may test higher; a maximum-fiber guarantee tells you the ceiling, not the content. State feed control officials can and do test products against their guarantees. Second, the numbers are as fed, so a kibble at roughly 10% moisture and a fresh food at roughly 75% cannot be compared line for line — that needs a dry-matter or energy basis.

We publish each food's guaranteed analysis as the manufacturer states it, label the basis, and keep every derived number in a separate column.

A typical analysis reports what the food actually contains on average, rather than the floor or ceiling the manufacturer is willing to guarantee. Some brands publish typical values in addition to the required guaranteed analysis, often on a dry-matter basis; Hill's does this most consistently.

Typical values are usually the more useful number, because a guarantee of "minimum 26% protein" and an average of 29% describe the same food very differently. But they are voluntary and not enforceable, and they are not directly comparable to another brand's guarantees. Mixing the two silently — comparing one brand's typical protein against another's guaranteed minimum — flatters whichever brand published typical values.

We never mix them. Each food in the database records its basis as either guaranteed or dry_matter_typical, every table says which it is, and typical dry-matter values are passed through without a second conversion.

Moisture is the water in the food, guaranteed as a maximum percentage. It is the single most important number on the guaranteed analysis, because every other percentage is diluted by it.

In our database, dried formats sit around 5–15% moisture, kibble around 10%, canned food around 78%, and cooked fresh recipes between about 69% and 82%. A canned food at 8% protein and 78% moisture is 36.4% protein once the water is removed — comparable to a 32% kibble, not a quarter of it. This is why as-fed protein percentages across formats are meaningless without conversion.

Moisture also drives cost. When roughly three-quarters of what you buy is water, price per pound describes the water as much as the food, which is why we compare on cost per 1,000 kcal instead.

Crude protein is not measured directly. A laboratory measures the nitrogen in a sample (by Kjeldahl digestion or combustion analysis) and multiplies it by a fixed factor of 6.25, on the assumption that protein averages 16% nitrogen. That is what "crude" means: an indirect estimate, not an analysis of the protein itself.

Three consequences. The number says nothing about amino acid profile — a food can meet a protein guarantee while being short on a specific essential amino acid, which is why AAFCO's nutrient profiles set individual amino acid minimums. It says nothing about digestibility, so protein from feathers and protein from muscle meat count the same. And any nitrogen counts, including non-protein nitrogen, which is the loophole melamine adulteration exploited in the 2007 pet food recalls.

It remains a useful comparative number once converted to a dry-matter or per-1,000-kcal basis. It is not a quality score.

Crude fat is determined by extracting the sample with a solvent and weighing what dissolves. Like crude protein, it is a method-defined number rather than a direct measure of dietary fat, and it does not distinguish fat sources or fatty acid profile. Labels that make a claim about omega-3 or omega-6 fatty acids must guarantee those separately; we record those extra guarantees where brands publish them.

Fat carries roughly two and a half times the energy of protein or carbohydrate per gram, so it drives calorie density more than anything else on the panel. Two foods with the same protein guarantee and different fat guarantees will feed out in different portion sizes.

Because it is a minimum, the real figure is usually higher. Compare fat on a dry-matter or per-1,000-kcal basis — never as-fed across formats.

Crude fiber is what survives boiling a sample in acid and then in alkali. The method mostly recovers cellulose and lignin, and misses much of the soluble and fermentable fiber a food contains. Total dietary fiber, the measure used in nutrition research, is typically higher — often substantially so.

So a low crude fiber guarantee does not mean a low-fiber food, and two foods with identical crude fiber guarantees can behave quite differently. It is a maximum, which means the label is promising a ceiling rather than describing a level.

We use crude fiber the way the label states it, and in our carbohydrate estimate, where it is one of the terms subtracted from 100. Because crude fiber undercounts total fiber, the leftover carbohydrate figure is an estimate with error in both directions, and we mark it as one.

Ash is what remains when a food sample is incinerated: calcium, phosphorus, sodium, magnesium and the other minerals, plus whatever mineral content comes from bone. It is not a filler and it is not an ingredient; it is a measurement.

European labels routinely declare crude ash. US labels are not required to, and most do not. That matters here because ash is one of the terms in the standard carbohydrate estimate, so when it is absent we have to assume a value. We use 8% for dry formats and 2.5% for wet and fresh ones — foods above 50% moisture — which are the conventional working assumptions. Where a brand does publish ash, the estimate is still marked as an estimate.

Foods high in bone or bone meal carry more ash. That is a formulation fact, not a defect, but it does mean a higher assumed ash figure would be more accurate for some raw and bone-inclusive products than the flat assumption we apply.

US dog food labels must state calorie content, expressed as kilocalories of metabolizable energy per kilogram of food as fed, plus a per-unit figure in a familiar household measure — per cup, per can, per patty. AAFCO's model regulations made this mandatory on all dog and cat food labels in 2017; before that it appeared mainly on foods making "light" or weight-related claims.

The statement is the most useful number on the package. It converts a feeding guideline into arithmetic: daily calories divided by kcal per cup is the portion. It also makes formats comparable, because a kilocalorie is a kilocalorie whether it arrives as kibble or as refrigerated fresh food.

The energy figure may be derived by calculation or by feeding trials, and labels do not always say which. We record kcal per kilogram and the per-unit figure exactly as published, and leave the field empty when a manufacturer does not publish it.

Ingredients are listed in descending order by weight as the food is formulated, before cooking. Splitting takes advantage of that ordering. A recipe containing a large amount of pea material can list "peas," "pea protein," "pea fiber" and "pea starch" as four separate entries, each smaller than a single combined "peas" entry would be, so a meat ingredient appears first instead of second or third. The same is done with rice (brewers rice, rice bran, rice flour) and potato.

Nothing about this is illegal — the fractions are genuinely different ingredients with different compositions. It simply means ingredient order cannot be read as a proportion.

The correction is to stop reading the list as a ranking of quantity and read the analysis instead. Protein, fat and estimated carbohydrate on a dry-matter or per-1,000-kcal basis describe what is actually in the food; the ingredient list describes what went in and in what order.

AAFCO defines these terms precisely. Meat is the clean flesh of slaughtered mammals — striated muscle with the fat, skin, sinew, nerve and blood vessels that normally accompany it. Meat meal is the rendered product of mammal tissues, excluding blood, hair, hoof, horn, hide trimmings, manure and stomach contents. Meat and bone meal is the same process with bone included. Species-named versions — chicken meal, lamb meal — identify what the meal was made from.

The practical difference is water. Fresh meat is roughly 70% water, and it is weighed with that water when the ingredient list is ordered. A meal is around 10% moisture. So "deboned chicken" listed first can contribute less protein to the finished food than "chicken meal" listed third. Neither is inherently the better ingredient; a meal is a concentrate, not a downgrade.

What the term does not tell you is the grade of the raw material. It tells you the species and the process.

AAFCO's definitions again. Meat by-products are the non-rendered clean parts of slaughtered mammals other than meat: lungs, spleen, kidneys, brain, liver, blood, bone, and stomachs and intestines freed of their contents. Hair, horns, teeth and hooves are excluded. Poultry by-product meal is the rendered, ground clean parts of poultry carcasses — necks, feet, undeveloped eggs, intestines — exclusive of feathers except in trace amounts unavoidable in processing.

The word carries more stigma than the definition supports. Organ tissue is nutritionally dense, and much of what is excluded from human retail cuts is eaten routinely elsewhere. What is fair to say is that by-product ingredients are more variable in composition than a single named muscle meat, and that an unnamed by-product meal does not tell you the species.

We record ingredients as printed and do not grade them. Whether a by-product meal belongs in your dog's bowl is a judgement about variability and species labelling, not a nutritional verdict we can make for you.

AAFCO ingredient definitions come in named and generic versions. "Chicken meal," "lamb meal" and "chicken fat" name a species. "Meat meal," "meat and bone meal," "animal fat" and "animal digest" do not — the generic definitions permit material from more than one mammal species, and the mix can vary between production runs.

This matters in two practical ways. If you are avoiding a particular species for any reason, a generic ingredient makes that impossible to verify from the label. And an unnamed source is less traceable: if you want to know what changed after a reformulation, a named ingredient gives you something to compare.

It is not, on its own, a quality signal. Named ingredients can be low grade and generic ones can be perfectly good. It is a transparency signal — the same thing we track when a brand declines to publish its guaranteed analysis or calorie data at all.

Ingredients appear in descending order by weight as formulated, so the first few entries usually account for most of the recipe. Reading them tells you what kind of food you are holding: meat-first or grain-first, legume-heavy or not, one protein or several.

What it cannot tell you is how much. Order is by pre-cooking weight, so a fresh meat weighed with its water outranks a dried meal that contributes more protein. Ingredient splitting pushes single commodities down the list in fractions. And the list never reveals the gap between first and second — the top ingredient might be 40% of the recipe or 12%.

Treat the first five ingredients as a description of style, then check the numbers. Protein, fat and estimated carbohydrate on a dry-matter basis, plus calorie density, describe what the dog actually eats. We publish the first ingredients for every food alongside the analysis so you can hold the two against each other.

Grain-free means a recipe contains no cereal grains. It does not mean low carbohydrate. Extruded kibble needs starch to form and expand, so the grains are generally replaced with peas, lentils, chickpeas, potatoes or sweet potatoes, and the finished carbohydrate content is often similar. Our estimated carbohydrate figures let you check that on any given food rather than assume it.

The term has no definition in AAFCO's model regulations and no established nutritional meaning. Dogs have no dietary requirement for grain, and equally no requirement to avoid it. A dog with a diagnosed sensitivity to a specific ingredient is a veterinary matter, not a label-shopping one.

We record grain-free as a flag on each food because people search on it, and because of its role in the FDA's DCM investigation — which reported an association in adverse-event reports and did not establish that grain-free diets cause disease. We draw no health conclusion from the flag.

"Novel" only means unfamiliar to that particular animal. Venison is novel to a dog raised on chicken and entirely unremarkable to one that has eaten venison for years. There is no regulatory definition and no list of novel proteins; the word on a package is a marketing description of an uncommon ingredient.

Veterinary elimination-diet work uses proteins the patient has not encountered, or hydrolysed proteins broken into fragments. That is a clinical process with a defined protocol, run by a veterinarian, and it is not something a retail label can carry out for you.

One documented caveat is worth knowing: published testing of over-the-counter novel-protein and limited-ingredient products has repeatedly found animal proteins that were not declared on the label. That is a reason to treat retail novelty claims as approximate.

We list ingredients as printed and make no claim about what any diet does for any condition.

A limited-ingredient product usually means a single named animal protein, a single main carbohydrate source, a fat, and the vitamin and mineral premix that makes the food complete. Nothing in AAFCO's model regulations defines the term, sets a maximum ingredient count, or requires anything in particular to be left out. Two products both labelled limited-ingredient can differ substantially.

The practical value is legibility: a shorter list is easier to compare against another food, and easier to describe to a veterinarian. Several of the foods in our database are labelled this way, and their ingredient lists are published in full on each food page.

Two cautions. A short list is not automatically a simple one — a "single protein" recipe can still contain fats or flavourings from other species. And published testing has found undeclared proteins in retail limited-ingredient products. If a diet is being used to investigate a suspected food reaction, that is a veterinary process, not a shopping decision.

As-fed percentages are diluted by whatever water the food contains, so they cannot be compared across formats. Dry-matter conversion removes the water from the arithmetic. DM% = 100 − moisture, then nutrient_DM = as-fed ÷ DM% × 100.

Worked example: a canned food guaranteeing 8% protein at 78% moisture has 22% dry matter, so 8 ÷ 22 × 100 = 36.4% protein on a dry-matter basis. Against a 32% kibble at 10% moisture — 35.6% dry matter — the canned food is slightly the higher-protein product, not a quarter of it, which is what the as-fed numbers suggest.

We convert every food this way. Where a brand publishes typical values already on a dry-matter basis, we pass them through rather than converting twice, and the basis is labelled on the page. Dry matter is the right basis for comparing composition; for comparing nutrient supply relative to calories, use grams per 1,000 kcal.

As fed is the food in the state the dog eats it. Every guaranteed analysis on a US label is as-fed, and so is the calorie content statement's kcal-per-kilogram figure.

Within one format, as-fed numbers compare fine — two kibbles at about 10% moisture are close enough that the water barely moves the comparison. Across formats they mislead badly, and the error runs in the direction that flatters dry food. A fresh recipe at 75% moisture has only a quarter of its weight available to carry nutrients, so its as-fed protein percentage is roughly a quarter of its dry-matter percentage.

This is also why feeding amounts differ so much: a dog eating a fresh diet eats several times the weight of food it would eat as kibble for the same calories. We show as-fed values because they are what the label says, and dry-matter and per-1,000-kcal values beside them because those are what compare.

Carbohydrate is not a required guarantee on US pet food labels, so it is estimated by subtracting everything else from 100. Nitrogen-free extract is the standard name for that remainder: mostly starch and sugars, plus the soluble fiber that crude fiber analysis misses.

We compute it as 100 − protein − fat − fiber − moisture − ash on an as-fed basis, using the label's guarantees. Ash is rarely published, so we assume 8% for dry formats and 2.5% for wet and fresh — foods above 50% moisture. Where the arithmetic returns zero or less, we publish nothing rather than a number.

Treat the result as an estimate and nothing more. It is built from minimums and maximums rather than measurements, and every error in those guarantees lands in the remainder. We mark it with a tilde and an asterisk everywhere it appears. If AAFCO's label modernization puts carbohydrate on packages, we will switch to declared values.

Not all the energy in a food reaches the animal. Metabolizable energy is what is left after subtracting the energy lost in feces and urine, and it is the basis on which pet food calorie content is declared and on which AAFCO's nutrient profiles are expressed.

ME can be determined two ways. It can be measured, by feeding the food to dogs and analysing intake and excreta. Or it can be calculated — AAFCO's modified Atwater factors assign 3.5 kcal per gram of protein, 8.5 per gram of fat and 3.5 per gram of carbohydrate, deliberately lower than the human food factors because pet foods are less digestible than the mixed human diet those were derived from.

Calculated and measured values for the same food can differ, and labels do not always state which method produced the number. We use the figure the manufacturer publishes, as published, and note when none is published at all.

Kilocalories per kilogram is the calorie content statement's primary figure, given on an as-fed basis. It answers a question no other label number does: how much energy is in this food per unit of weight you are buying.

The spread is enormous. Kibble in our database sits in the low thousands of kcal per kilogram; refrigerated fresh recipes, being mostly water, sit at a fraction of that. Per-cup and per-can figures are convenient but not comparable — cups vary with kibble density and scoop technique, cans vary in size — so kcal per kilogram is the number we build everything else on.

Two derived figures come from it. Grams per 1,000 kcal (nutrient% × 10,000 ÷ kcal/kg) puts nutrient supply on an energy basis. Cost per 1,000 kcal puts price there too. Where a manufacturer does not publish kcal per kilogram, neither figure can be computed, and we show the gap.

Dogs eat to meet an energy requirement, so what matters is how much of a nutrient arrives alongside each calorie. That is why AAFCO publishes its dog food nutrient profiles on an energy basis as well as a dry-matter one, and why we compute it for every food that publishes calorie data.

The formula is grams per 1,000 kcal = nutrient% × 10,000 ÷ kcal per kg. A food at 26% protein and 3,600 kcal/kg supplies about 72 g of protein per 1,000 kcal.

This basis catches something dry matter hides. A high-fat food is energy dense, so a dog eats less of it, so every other nutrient in it arrives in a smaller portion. Two foods can look similar on a dry-matter basis and deliver noticeably different protein per calorie. Where a label omits calorie content, this figure cannot be calculated and the column stays empty.

Resting energy requirement is the baseline: the energy an animal uses at rest, awake, not digesting a large meal, in a comfortable temperature. The standard veterinary estimate is RER = 70 × kg^0.75, where kg is body weight in kilograms.

The exponent matters. Energy needs do not scale linearly with body weight — a 40 kg dog does not need twice the calories of a 20 kg one — which is why per-pound feeding rules of thumb go wrong at both ends of the size range. Our feeding calculator uses this exact formula.

RER is not a feeding amount. It is multiplied by a factor reflecting activity, neuter status and life stage to give a daily estimate. And it is a population average: individual dogs vary by 25% or more in either direction, so any number derived from it is a starting point to be adjusted against body condition, not a prescription.

Maintenance energy requirement is the everyday number: what an animal needs to hold its weight while living a normal life. It is estimated as MER = RER × factor. Commonly used factors run from around 1.2 for weight loss and 1.4 for a less active or older dog, through roughly 1.6 for a typical neutered adult and 1.8 for an intact or active one, up to 2.5–3.0 for growing puppies. Our calculator defaults to 1.6.

These multipliers are population estimates drawn from veterinary nutrition texts, and the spread between individual dogs is wide — two dogs of the same weight and description can differ by a quarter or more. Working dogs, cold climates, gestation and lactation all sit outside the ordinary range.

Use the output as a starting portion, then let the dog's body condition over two to four weeks correct it. Dogs with medical conditions, pregnant or lactating dogs, and dogs at unusual weights need a veterinarian, not a calculator.

Body condition scoring rates an animal's fat cover by feel and sight rather than by the scale. The common veterinary version runs from 1 to 9, with the middle of the range treated as ideal; a 5-point version exists and is scored the same way in principle. Assessment looks at whether ribs can be felt without pressing, whether a waist is visible from above, and whether there is an abdominal tuck from the side.

It exists because weight alone is uninformative. Breeds and frames differ so widely that a number on a scale means little without knowing what that individual should carry, and owners are consistently poor at judging their own dogs by eye.

Body condition is the feedback loop for any feeding calculation. A calculator gives a starting portion; the dog's condition over the following weeks says whether it was right. Scoring is best learned from your veterinarian, who can show you the landmarks on your own dog. We do not score dogs or give feeding prescriptions.

AAFCO is an association of the officials who regulate animal feed — state departments of agriculture, federal agencies including the FDA, and international members. It is not a government agency, it does not test or approve foods, and it cannot enforce anything.

What it produces is model language: ingredient definitions, model labelling regulations, and the dog and cat food nutrient profiles. Individual states then adopt that language into their own feed laws, and state feed control officials enforce it. This is why a label can be non-compliant in one state and not another, and why "AAFCO approved" is a phrase that means nothing — no such approval exists.

What does exist is a nutritional adequacy statement, printed on the label, in which the manufacturer states which AAFCO profile the food meets and how that was substantiated. We quote that statement verbatim on every food page, because the exact wording is the claim.

The AAFCO Dog Food Nutrient Profiles list the nutrients a complete diet must supply and the levels it must supply them at — protein, fat, essential amino acids, fatty acids, minerals and vitamins. Most entries are minimums; a few carry maximums where excess is a concern.

There are two profiles. Growth and reproduction covers puppies and gestating or lactating females and is the more demanding of the two. Adult maintenance covers adult dogs not growing or reproducing. A food meeting the growth and reproduction profile can be labelled for all life stages, because that profile subsumes maintenance. Foods claiming growth must additionally state whether they are suitable for the growth of large-size dogs.

The profiles are stated on a dry-matter basis and per 1,000 kcal of metabolizable energy — which is why we compute both bases for every food. They set a floor for a complete diet. They do not describe an optimum, and meeting them is not a ranking.

Every dog food label must carry one. It takes one of three forms. Either the food is formulated to meet the AAFCO nutrient profiles for a named life stage; or animal feeding tests using AAFCO procedures substantiate that it provides complete and balanced nutrition for a named life stage; or it is intended for intermittent or supplemental feeding only, meaning it is not a complete diet.

A fourth variant exists for product families, where a food is stated to be comparable in nutritional adequacy to a product that passed a feeding trial.

The wording carries the whole claim, which is why we reproduce it word for word rather than summarising it. "For maintenance" and "for all life stages including growth of large size dogs" are different products of the same brand line. Where a manufacturer publishes no statement on a public page, we record that gap instead of inferring one.

This is the more common of the two substantiation routes and the weaker evidence level. The manufacturer demonstrates that the formula meets the relevant AAFCO nutrient profile, using ingredient nutrient databases, laboratory analysis of the finished product, or both. No dogs are involved.

What it establishes is that the nutrients are present at the required levels. What it does not establish is whether they are digestible and available to the animal, or whether dogs will eat and thrive on the food over months. Nutrient content and nutrient delivery are not the same thing — a mineral can be present and poorly absorbed.

This is not a criticism of the method; it is the standard route, and formulation science is well developed. It is simply a different claim from a feeding trial, and the label distinguishes them. Roughly speaking, most foods in our database use this route. We tag each food's evidence level from its verbatim statement.

When a label reads "animal feeding tests using AAFCO procedures substantiate that [product] provides complete and balanced nutrition for [life stage]," the finished food was fed to dogs under a defined protocol and they were monitored throughout. It is the stronger of the two substantiation routes, because it tests delivery rather than composition: dogs actually ate the food, and defined health measures were recorded.

Trials cost money and take months, so relatively few products carry this statement. In our database it appears on Hill's Science Diet products and on some JustFoodForDogs recipes; most other foods use the formulation route. That is a fact about evidence level, not a verdict about quality — a formulated food is not thereby deficient.

It is also worth knowing what the trial does not do. It is short, small and pass-or-fail; it does not compare foods, measure long-term outcomes, or establish that one diet is better than another.

The AAFCO adult maintenance protocol sets a floor, not a research standard. In outline: a minimum of eight dogs begin the test, the food is fed as the sole ration for 26 weeks, a veterinarian examines the dogs at the start and finish, dogs must not lose more than a set proportion of their starting body weight, and four blood measures are recorded at the end — hemoglobin, packed cell volume, serum alkaline phosphatase and serum albumin. Dogs removed for non-nutritional reasons must be accounted for. Separate protocols cover growth and gestation/lactation.

Read that list and its limits are obvious. Eight dogs is a small group. Six months is a short window against a decade-long lifespan. Four blood values are a narrow window on health, and there is no control diet and no comparison.

It is still meaningfully more than a spreadsheet: the food was eaten, in full, for months, without measurable harm on those parameters. That is the claim, and it is worth exactly what it says.

AAFCO recognises a small set of life stages, and the adequacy statement must name one. Growth covers puppies. Gestation/lactation covers breeding females. Maintenance covers adult dogs that are neither growing nor reproducing. All life stages means the food meets the growth and reproduction profile, which is the more demanding one and therefore also covers maintenance.

"Senior" is not an AAFCO life stage. There is no senior nutrient profile, so a food marketed for older dogs is substantiated for adult maintenance like any other; the same is true of "puppy" foods, which are substantiated for growth. Breed-size and "weight management" descriptors are likewise marketing segmentation sitting on top of one of the recognised stages.

Because an all-life-stages food is built to a growth profile, it carries nutrient levels aimed at puppies. That is a suitability question for a particular dog, not a defect. We record each food's stated life stage exactly as the label puts it.

Foods substantiated for growth or for all life stages must say which of two things is true: that they are suitable including the growth of large size dogs (70 lb or more as an adult), or except for that growth. You will see both phrasings in the statements we quote, sometimes differing between two recipes in the same product line.

The distinction turns on calcium. AAFCO's growth profile sets a maximum as well as a minimum, and the large-size-dog qualification requires the tighter ceiling, because calcium intake during the rapid growth phase of large-breed puppies is a recognised concern in veterinary nutrition — well enough recognised that the label carries the warning.

Practical reading: if you have a puppy likely to mature at 70 lb or more, the clause is the part of the statement to check, and it is a question for your veterinarian rather than for us. For an adult dog of any size it is not relevant.

"Complete and balanced" is a regulated claim, not a compliment. Complete means every nutrient known to be required is present; balanced means they are present in appropriate proportions to one another. The claim is always tied to a life stage and always backed by one of the two substantiation methods, both of which appear in the food's adequacy statement.

Its opposite on a label is "intended for intermittent or supplemental feeding only" — the required statement for products that are not complete diets. Toppers, mixers, treats and some raw products carry it. It is not a warning about safety; it means the product is not designed to be the entire ration, and feeding it as one would be.

The claim's limit is the phrase "known to be required." Profiles reflect current knowledge, revised as the science moves. A complete and balanced food is one that meets a published standard for a named population. That is a floor, not a ranking, and it is not a claim about any individual dog.

If a food does not meet an AAFCO nutrient profile and has not passed a feeding test, its label must say so, in the words "This product is intended for intermittent or supplemental feeding only." The sentence appears in the same place as a nutritional adequacy statement and is easy to skim past.

Products that carry it include treats, toppers and mixers, some raw and freeze-dried products, and single-ingredient items. There is nothing wrong with any of them — they are just not rations. A dog fed one exclusively over time is being fed an incomplete diet, and the label has said as much.

Some therapeutic veterinary diets also carry this statement, for a different reason: they are formulated for a clinical purpose under veterinary supervision rather than to a general profile. Those are prescribed, not chosen from a shelf.

Where a product in our database carries this statement rather than an adequacy claim, we quote it as it stands.

Kibble is the dominant format by volume: ground ingredients cooked and formed into dry pellets, dried to roughly 10% moisture, then usually coated with fats and palatants to make them appealing. Nearly all kibble is extruded; a smaller amount is baked or cold-formed.

Its advantages are structural. Low moisture makes it shelf-stable and cheap to ship, the cooking step is a validated kill step for pathogens, and calorie density is high, so portions are small. Cost per 1,000 kcal is lower than any other format in our database, generally by a wide margin.

Its constraint is also structural: forming and expanding a pellet requires starch, so extruded foods carry a meaningful carbohydrate load whether or not they contain grain. Our estimated carbohydrate column shows roughly where a given kibble lands. Fat coatings can oxidise over time, which is what the best-by date and the resealed bag are for.

In an extruder, a dough of ground ingredients is cooked under heat, pressure and mechanical shear, then forced through a shaped die. The sudden pressure drop at the die flashes off water and puffs the dough into a pellet, which is cut to length, dried, and typically coated with fat and palatants.

Three things follow. The process needs starch — starch is what gelatinises and holds the expanded structure — so extruded foods contain carbohydrate by design; grain-free recipes substitute legumes or potatoes rather than removing the requirement. The heat is a pathogen kill step, which is why extruded food is handled casually in a way raw food is not. And heat degrades some vitamins, which is why vitamin premixes are dosed with processing losses in mind and sometimes applied after cooking.

Extrusion is neither a virtue nor a defect. It is a manufacturing method with known consequences for composition and handling.

Fresh cooked foods are gently cooked rather than extruded or retorted, then chilled or frozen and shipped to the customer, usually on a subscription with pre-portioned packs. Cooking is a kill step, so handling is ordinary refrigerated-food handling rather than raw-food handling.

The defining number is moisture. Fresh recipes in our database run from about 69% to 82% water, which changes everything downstream: as-fed guarantees look low until converted to dry matter, calorie density per kilogram is a fraction of kibble's, portions by weight are several times larger, and cost per 1,000 kcal is the highest of any format we track.

"Human grade" is a separate, defined claim about ingredients and facility, not a synonym for fresh. Adequacy is a separate question again, answered by the label's AAFCO statement — many fresh recipes are formulated to meet a profile, and a few are substantiated by feeding tests. We record which, per recipe.

Air-drying removes water gradually using warm circulating air instead of the pressure and shear of an extruder. The finished product is dense and shelf-stable — around 15% moisture in the products we track — and is usually fed in smaller portions by weight than kibble because it is calorie dense.

Because there is no expansion step, air-dried recipes do not need starch to hold their structure, so carbohydrate content is a formulation choice rather than a process requirement. Cost per 1,000 kcal sits between kibble and fresh.

One thing to check rather than assume: air-drying is a drying process, not automatically a validated pathogen kill step. Some air-dried products are made from cooked material, some are closer to raw, and manufacturers use different combinations of heat, time and post-processing. If that distinction matters to your household, the manufacturer is the place to ask. Adequacy, as ever, is answered by the AAFCO statement on the label, which we quote in full.

Freeze-drying freezes the food and then places it under vacuum, so the ice sublimates — passes directly from solid to vapour — without a liquid stage. Very little heat is involved, which preserves structure and volatile compounds and makes the process popular for raw diets, toppers and treats. The result is light, low in moisture and shelf-stable until opened; most products are rehydrated before feeding.

The important caveat is that freeze-drying is a dehydration process, not by itself a validated pathogen kill step. A freeze-dried raw product is still a raw product, and the FDA's handling guidance for raw pet food applies to it. Some manufacturers add a separate validated treatment, such as high-pressure processing; that is a manufacturer-specific claim to verify, not a property of the format.

Many freeze-dried products are treats or toppers carrying an "intermittent or supplemental feeding only" statement rather than a complete-diet claim. Check the adequacy statement before feeding one as a ration.

Raw products are built on uncooked meat, organ and often bone, sometimes with vegetables and a vitamin-mineral premix. Some are complete and balanced under an AAFCO life-stage claim; many carry the "intermittent or supplemental feeding only" statement instead. The adequacy statement is the place to check.

The FDA's published position is that it does not recommend feeding raw meat diets to pets. It cites its own sampling of pet foods, in which raw pet food was more likely than other pet food types to test positive for Salmonella and Listeria monocytogenes. The agency's stated concern covers both the animal and the people in the household, since bacteria can be present in the food, on the packaging, in the bowl and in the dog's stool. FDA guidance is handling guidance — refrigerated thawing, hand and surface washing, careful bowl hygiene — not a prohibition; raw pet food is legally sold. Raw and freeze-dried raw products do appear in FDA recall notices for pathogen contamination.

We publish raw products with the same label data and recall history as everything else. Whether raw suits a particular household, especially one with immunocompromised members, is a conversation with a veterinarian.

Shelf-stability is a storage property, not a nutritional one, and it is reached by two routes. The first is removing water: kibble, air-dried and freeze-dried products are dry enough that spoilage organisms cannot grow. The second is retorting — sealing wet food in a can or pouch and heating it under pressure until it is commercially sterile. That is why canned food keeps for years unopened and needs refrigeration within days after opening.

Refrigerated and frozen fresh foods are not shelf-stable, which is why they ship cold and cost more to distribute.

Shelf-stable is not a defined regulatory term for pet food, so it describes how a product behaves rather than promising anything. Storage still matters after opening: fat coatings oxidise, dry food picks up moisture from humid air, and best-by dates assume a sealed package. Keeping the original bag — with its lot code — is the practical habit.

Rendering applies heat to animal tissue and separates it into three streams: fat, water, and protein solids. The solids are dried and ground into meals — chicken meal, lamb meal, meat and bone meal — and the fat becomes ingredients such as chicken fat or generic animal fat. It is a long-established, regulated industrial process, and it is how most of the protein-dense ingredients in dry pet food are produced.

Two neutral facts about it. Rendered meals are concentrated: with most of the water removed, they carry far more protein per unit weight than fresh meat, which is why their position in an ingredient list understates their contribution. And the process is defined by what goes into it — AAFCO's definitions specify which tissues are excluded from each meal, but not the grade of the raw material.

So the ingredient name tells you the species (if named) and the process. It does not tell you the quality of the inputs, and no label term does.

Dilated cardiomyopathy (DCM) is a disease of the heart muscle. In July 2018 the US Food and Drug Administration announced it was investigating reports of DCM in dogs eating diets that frequently listed peas, lentils, other legume seeds or potatoes among the main ingredients — many, though not all, marketed as grain-free.

What the FDA published: counts of the adverse-event reports it had received and, in a June 2019 update, the brand names appearing most often in them. What the FDA said about that data: adverse-event reports are voluntarily submitted and unverified, the number of reports for a brand reflects how many dogs eat it as well as anything else, and the agency had not established a causal relationship between these diets and DCM. It did not order a recall and did not advise the public to stop feeding any particular product.

In December 2022 the FDA said it did not intend to issue further public updates unless there was meaningful new scientific information, and repeated that no causal link had been established. Research continues in the veterinary literature and the question is not settled. We record grain-free status as a label fact and draw no health conclusion from it; diet questions about an individual dog belong with a veterinarian.

A recall is a firm's removal or correction of a product already on the market. The overwhelming majority are voluntary: the company initiates them, sometimes on its own testing, sometimes after an FDA request. Under the Food Safety Modernization Act the FDA also has authority to order a mandatory recall of food, including pet food, if a company will not act — a power used rarely, because the request usually suffices.

"Voluntary" therefore says less about severity than people assume. A voluntary recall can be a Class I event affecting many lots, and a mandated one is unusual mainly because it means a company refused.

What is worth reading in a recall notice: the reason, the classification, the lot codes and dates affected, and whether illnesses were reported. A single-lot recall caught by routine testing and a multi-brand plant contamination are different events that share a word. Our recall tracker records each event with its source, and each brand page carries its full history.

The FDA distinguishes a market withdrawal from a recall. A withdrawal covers a product that involves a minor violation not subject to FDA legal action, or no violation at all — normal stock rotation, a packaging or labelling change, a quality complaint the firm chooses to act on. A related term, stock recovery, applies when the product never left the firm's control.

The distinction matters when reading brand histories. A withdrawal is not a recall and does not appear in the FDA's enforcement report with a class attached, so counting withdrawals alongside recalls overstates a brand's record, while ignoring a genuine recall because a company's press release calls it a "withdrawal" understates it. The words in a company statement are not always the regulatory category.

We track recall events, record what the FDA record says, and link the source on every one so the classification can be checked rather than taken from us.

The FDA assigns a class to each recall after the firm initiates it, and publishes it in the weekly Enforcement Report. The definitions come from the regulations.

Class I: there is a reasonable probability that use of, or exposure to, the violative product will cause serious adverse health consequences or death. Class II: use may cause temporary or medically reversible adverse health consequences, or the probability of serious consequences is remote. Class III: use is not likely to cause adverse health consequences — often a labelling or regulatory violation rather than a hazard.

Two points follow. The class is the FDA's assessment, not the company's, so it is a more reliable severity signal than a press release. And a Class I designation describes potential, not confirmed harm: many Class I pet food recalls are precautionary, triggered by a positive pathogen test with no reported illnesses.

Where the class is on record, we record it, and every event in our tracker links to its source.

A large share of pet food is made by contract manufacturers producing for several brands at once. The name on the bag is the marketer; the plant may belong to someone else entirely, and a brand may use different plants for different formats — kibble at one, canned at another.

This is ordinary manufacturing practice, and it says nothing about quality on its own. It matters for two reasons. First, recall scope: when a plant has a contamination problem, every brand it produces for can be affected at once, which is why some recall events span brand names that look unrelated on a shelf. Second, control: a brand that owns its plants controls its own process and testing; a brand that buys capacity is dependent on someone else's.

Brands rarely publicise these relationships, and we do not infer them. Where a manufacturing relationship is on the public record we note it on the brand profile, because it is part of reading a recall history properly.

A lot code identifies the production batch: which plant, which run, which day. It is normally printed or stamped near the best-by date on the bottom of a bag, the end of a can, or the seal of a pouch, and the format varies by manufacturer.

Recalls are almost always limited to specific lots. A notice will list the product name, size, best-by dates and lot codes affected, and everything outside that list is unaffected. Without the code from your package you cannot tell whether a recall applies to you, which is the whole practical reason the code is worth knowing about.

The habit that makes this work: keep the original packaging until the food is finished, or photograph the code panel before decanting into a storage bin. It also matters for reporting — if you want to report a problem to the FDA or the manufacturer, the lot code is the first thing either will ask for.

You are buying calories, not weight. When one food is 10% water and another is 78%, price per pound compares the water as much as the food, and the dry format always looks worse than it is.

The fix is arithmetic: cost per 1,000 kcal = price ÷ total package kcal × 1,000, where total package kcal is the package weight in kilograms multiplied by the label's kcal per kilogram. Both inputs are printed on the package, so this can be done in a shop.

The result is comparable across every format — kibble, canned, air-dried, fresh, raw — and the spread between formats is usually several-fold, wide enough that anyone switching formats should know the multiple before committing. Our calculator takes calorie density from each food's verified label data so you only need to enter the price and size you are actually paying.

Cost per 1,000 kcal compares foods; cost per day tells you what a switch actually does to your spending. The conversion is one step: take the dog's estimated daily calories, divide by 1,000, and multiply by the food's cost per 1,000 kcal.

It is dog-specific by design, because the same food costs very different amounts to feed a 10 lb dog and an 80 lb one — and that difference is the reason format comparisons that quote a single price are useless. A premium worth paying for a small dog can be a different decision entirely for a large one.

Two cautions. The daily calorie figure is an estimate with wide individual variation, so treat the result as a planning number. And subscription pricing frequently includes an introductory discount that ends; the number that matters is the ongoing one. Our cost tool reports a per-day figure alongside the per-calorie one.