You need a Nutrition Facts label. Do you have to send your product to a lab, or can you calculate the numbers from your recipe?
For most small food businesses, the answer is: you can calculate them. FDA doesn’t require lab testing for Nutrition Facts labels. In FDA’s words, “the source of the data used to calculate nutrition label values is the prerogative of the manufacturer.”
What FDA does require is that the numbers on your label be accurate. The method is your choice. The responsibility is yours either way.
This article explains how each method works, where each one falls short, and how to decide which your product needs.
The Short Answer
| Database (recipe) analysis | Laboratory testing | |
| How it works | Calculates nutrients from your ingredients, quantities, and finished yield | Chemically measures samples of your finished product |
| Required by FDA? | No | No |
| Typical cost | Low; free tools exist | A full nutrition panel runs about $1,656 per sample at one major U.S. lab (2026 price) |
| Speed | Minutes | Days to weeks |
| Recipe changes | Recalculate instantly | Make the product, sample it, and test again |
| Captures processing effects | Only what you enter (moisture loss, processing loss) | Yes, in the samples tested |
| Captures batch-to-batch variation | No | Only if you test several representative lots |
| Can determine added sugars | Yes, from the recipe | Not when a food also has naturally occurring sugars |
| Guarantees compliance | No | No |
What Is Database Nutrition Analysis?
Database analysis calculates a food’s nutrients from its recipe instead of measuring the finished product:
| Step | What happens |
| 1 | Enter each ingredient and its weight |
| 2 | Match each ingredient to nutrient data (a government database, supplier specifications, or lab data) |
| 3 | Add up the nutrients for the whole batch |
| 4 | Adjust for moisture loss (water that evaporates during baking or cooking) |
| 5 | Adjust for processing loss (finished product lost to trimming, crumbs, or pan residue) |
| 6 | Divide by the finished weight to get nutrients per serving |
| 7 | Apply FDA rounding rules |
| 8 | Generate the Nutrition Facts label |
FDA’s guidance on nutrition databases describes this approach directly: software that calculates “label values derived from the nutrient content of ingredients that comprise a product’s recipe, while taking into account nutrient losses during processing.”
What Is Laboratory Testing?
A lab analyzes samples of your finished product. A full Nutrition Facts panel usually includes tests for moisture, protein, fat (with a fatty acid profile for saturated and trans fat), ash, dietary fiber, sugars, cholesterol, sodium and other minerals, and vitamin D. Carbohydrate and calories are then calculated from those results.
The result tells you what was in the samples you sent.
Does FDA Prefer Lab Testing?
Sort of, and this nuance matters. FDA’s database guidance says the agency’s “policy recommends that the nutrient values for labeling be based on product composition, as determined by laboratory analysis of each nutrient.”
But the same guidance:
- says the data source is the manufacturer’s choice,
- treats ingredient and recipe databases as legitimate tools and explains how to build and maintain them, and
- says submitting a database to FDA for review is voluntary.
So both of these statements are wrong:
- “FDA requires lab testing for every Nutrition Facts label.”
- “Database analysis is always as accurate as lab testing.”
The accurate version: database analysis is allowed and practical for most formulated foods, and you’re responsible for the accuracy of the result.
How FDA Checks Your Label
FDA doesn’t ask how you got your numbers. It tests the product.
Under 21 CFR 101.9(g), FDA collects 12 consumer units, one from each of 12 randomly chosen shipping cases in a lot, combines them into a single composite sample, and analyzes it. The result is compared with your label:
| Nutrient group | Examples | The lab result must be |
| Added nutrients (Class I) | Vitamins, minerals, protein, or fiber added to fortified foods | At least 100% of the declared value |
| Naturally occurring nutrients (Class II) | Protein, carbohydrate, fiber, vitamins, and minerals naturally in the ingredients | At least 80% of the declared value |
| Nutrients to limit | Calories, total sugars, added sugars (when all sugars are added), total fat, saturated fat, trans fat, cholesterol, sodium | No more than 120% of the declared value |
Reasonable amounts above the declared value for Class I and Class II nutrients, or below it for the nutrients to limit, are acceptable under good manufacturing practice.
This is why method matters less than accuracy. A calculated label that matches your product passes. A lab-tested label from one unrepresentative sample can fail.
Lab Testing Can’t Tell You Your Added Sugars
This surprises many people. When a food contains both added sugars and naturally occurring sugars (for example, granola with honey and dried fruit), no lab test can separate them.
For those foods, FDA requires manufacturers to keep records that support the added sugars declaration, such as recipes and ingredient specifications, and to keep them for at least two years after the food is introduced into commerce. In practice, that means even a lab-tested label relies on recipe-based calculation for added sugars.
FDA has similar record-keeping rules for a few other nutrients that can’t be measured directly, including some types of dietary fiber.
A Worked Example: Granola
Here’s how a database calculation handles a 7,595 g batch of granola:
| Stage | Weight | What changed |
| Ingredients weighed | 7,595 g | |
| After baking and cooling | 7,250 g | 345 g of water evaporated (4.5% moisture loss) |
| Usable finished product | 7,100 g | 150 g lost to crumbs and pan residue (2.1% processing loss) |
The two losses work differently:
- Moisture loss removes water only. The nutrients stay in the granola and become more concentrated per gram.
- Processing loss removes finished granola, nutrients and all. It reduces the number of servings but doesn’t change the nutrients per gram.
If you skip the moisture loss and assume the granola still weighs 7,595 g, every nutrient per serving comes out about 4.5% too low. That’s an error in the direction FDA cares about most for calories, fat, and sodium.
For a 55 g (1/2 cup) serving, the calculation gives 59 mg sodium, declared as 60 mg after rounding.
The Quality of Your Data Matters Most
A database calculation is only as good as the ingredient data and recipe you enter.
Match the Right Ingredient
| Your ingredient | Wrong match | Problem |
| Whole milk | 2% milk | Fat and calories too low |
| Sweetened dried cranberries | Unsweetened dried cranberries | Sugars far too low |
| Salted butter | Unsalted butter | Sodium too low |
| Your supplier’s semisweet chocolate chips | Generic “chocolate chips” | Fat, sugar, and fiber may all differ |
USDA FoodData Central is an excellent source, but a search for “milk” or “chocolate” returns dozens of entries with different fat levels, processing, and data types. The size of a database matters less than how well each entry matches what you actually buy.
Use Supplier Data for Commercial Ingredients
For purchased ingredients like chocolate chips, flavored syrups, or seasoning blends, your supplier’s nutrition specification is often the best data available. It describes the exact product you use, not an average.
Make Sure the Recipe Matches Production
The best database can’t fix a wrong recipe. If production uses 12 pounds of sugar but your nutrition recipe says 10, the label is wrong. Database analysis works best with:
- standardized formulas
- ingredients weighed, not added by eye
- batch records
- measured finished yields
- a process for updating the label when the recipe or a supplier changes
When Database Analysis Works Well
Database analysis is a strong fit for formulated foods where the recipe is known, ingredients are weighed, and the finished yield can be measured. That covers most small-business products:
- cookies, brownies, cakes, and breads
- granola, snack mixes, and energy bars
- chocolate and candy
- sauces, salsa, and dressings
- many beverages
When Database Analysis Gets Harder
Some processes change the food in ways a recipe can’t easily predict:
| Process | Why the recipe isn’t enough |
| Frying | Chips lose water and absorb an unknown amount of oil. Most of the oil in the fryer doesn’t end up in the product, so you can’t count it all. |
| Draining | Foods packed in brine, syrup, or oil may be labeled on a drained basis when the liquid isn’t usually eaten. Calculating from everything in the container overstates some nutrients. |
| Fermentation | Microorganisms consume sugars and produce acids, alcohol, and gases. A pre-fermentation recipe may overstate sugars. |
| Heat-sensitive nutrients | Some vitamins break down during cooking and storage. |
| Uneven distribution | Salt toppings, coatings, and inclusions may not spread evenly through the batch. |
| Variable agricultural ingredients | Tomatoes, fruit, and nuts vary by variety, season, region, and supplier. |
These don’t rule out database analysis. They mean you need good process data, such as measured fat absorption for a fried product, or a lab test to fill the gap.
One Lab Test Isn’t “The Truth”
Suppose your granola calculation shows 59 mg sodium per serving, and you test three production lots:
| Sample | Sodium per serving |
| Calculated | 59 mg |
| Lot 1 | 54 mg |
| Lot 2 | 61 mg |
| Lot 3 | 66 mg |
If you’d sent only Lot 1, you might have labeled 55 mg. If FDA later tested a lot like Lot 3, it would find 66 mg against a 55 mg declaration, exactly 120%, with no room for error.
A lab result describes the samples tested. It doesn’t remove variation in future batches. If you test, test more than one lot, and consider ingredient lots, suppliers, seasons, and production dates. A salsa made with July tomatoes from one farm and September tomatoes from another can differ more than you’d expect.
Claims Near a Threshold Need Extra Care
Nutrient content claims have fixed limits. A salsa can be called “low sodium” only if it has 140 mg or less per 2 tbsp and per 50 g.
If your calculation shows 138 mg, you have a 2 mg cushion. A different tomato supplier or a heavier hand with salt could push a batch over. If it shows 60 mg, you have plenty of room.
When a claim depends on a value close to its limit, lab testing several lots is cheap insurance.
Use Both: Calculate First, Test to Validate
You don’t have to choose one. A practical approach for growing manufacturers:
- Build the recipe in nutrition software and calculate the label.
- Start production.
- Send representative samples from several lots to a lab.
- Compare lab results with the calculated values.
- Investigate meaningful differences.
- Update ingredient data or loss percentages if needed.
FDA’s database guidance calls an ongoing comparison of lab results with database calculations critical to keeping a database accurate.
If you make 30 products, you don’t need to test all 30. Test the ones with difficult processes, claims near thresholds, or highly variable ingredients, and rely on calculation for the rest.
Targeted Tests Cost Much Less
You don’t always need a full panel. If the uncertainty in a fried snack is oil absorption, test fat and moisture. If you make a sodium claim, test sodium. Individual tests like protein or moisture can cost under $50 each, while fiber, vitamin D, and fat with a fatty acid profile cost several hundred dollars each.
When Lab and Calculation Disagree
Don’t assume one is wrong. Investigate:
| Possible cause | What to check |
| Wrong ingredient match | Does each database entry match what you actually buy? |
| Supplier differences | Does your supplier’s product differ from the generic entry? |
| Moisture loss | Is the finished product drier than your loss percentage assumes? |
| Fat absorption | Did frying add more fat than expected? |
| Recipe doesn’t match production | Are batch records consistent with the recipe in your software? |
| Uneven distribution | Are salt, toppings, or inclusions concentrated in some units? |
| Sampling | Were the samples representative of normal production? |
| Lab method | Was the method suitable for your food? All analytical methods have some uncertainty. |
A disagreement can teach you something about your process you didn’t know.
Choosing a Lab
- Choose a lab experienced in food nutrition analysis.
- Ask whether it uses AOAC Official Methods, which FDA recommends.
- Look for ISO/IEC 17025 accreditation.
- Ask whether its methods suit your food. Chocolate, high-fiber products, and high-fat products can all need specific methods.
Don’t choose on price alone.
Decision Guide
Database analysis is usually enough if:
☐ You have a standardized recipe and weigh ingredients ☐ Good nutrient data exist for your ingredients ☐ You can measure moisture loss and processing loss ☐ Production is consistent ☐ You’re developing or comparing several formulations ☐ You aren’t making claims close to a regulatory limit
Consider lab testing if:
☐ Your product is fried, fermented, or drained ☐ Your ingredients vary a lot by season or supplier ☐ A claim depends on a value near its limit ☐ A retailer or customer requires lab documentation ☐ Your calculated values look questionable
Consider both if:
☐ You have many products and want periodic validation ☐ Some nutrients are predictable and others aren’t ☐ You want low-cost formulation work plus independent verification
How FreeFoodLabels.com Fits In
FreeFoodLabels.com uses database analysis. You enter your recipe, and it calculates the nutrition from your formulation:
| You enter | FreeFoodLabels.com does |
| Ingredients and quantities | Calculates the nutrients in the recipe |
| Moisture loss and processing loss, as separate fields | Adjusts the finished yield correctly for each |
| Serving size and servings per container | Calculates per-serving values and applies FDA rounding |
| Generates the Nutrition Facts label |
Because the label comes from your recipe, you can change an ingredient, quantity, or loss percentage and recalculate in seconds. That makes it useful for product development as well as final labels. You can also keep the recipe records FDA expects for added sugars.
You can create and download a label free, with no account. Free labels include a small “FreeFoodLabels.com” line at the bottom. A $20 recipe unlock removes it and keeps the recipe editable with no time limit.
When a product needs lab validation, FreeFoodLabels.com gives you the calculated starting point to compare against.
Calculate your Nutrition Facts for free
Frequently Asked Questions
Does FDA require lab testing for Nutrition Facts labels?
No. FDA says the source of nutrition data is the manufacturer’s choice. The manufacturer is responsible for making sure the declared values are accurate.
Can I calculate Nutrition Facts from a recipe?
Yes. FDA’s database guidance describes recipe-based software that calculates label values from ingredient nutrient data while accounting for processing losses.
Does FDA approve nutrition software?
No. FDA doesn’t certify nutrition software or labels. Using a commercial program doesn’t mean FDA has reviewed your numbers.
Is lab testing more accurate than database analysis?
Not automatically. Lab testing measures real samples, which helps when processing is hard to predict. But results depend on representative sampling and appropriate methods. A careful recipe calculation can be just as reliable for most formulated foods.
Can a lab measure added sugars?
Not when a food contains both added and naturally occurring sugars. FDA requires recipe-based records to support the added sugars declaration in that case.
How accurate does my label have to be?
When FDA tests a product, naturally occurring nutrients like protein and fiber must be at least 80% of the declared value, added vitamins and minerals at least 100%, and calories, sugars, fat, and sodium no more than 120%.
What if my food loses weight during cooking?
Account for it. Moisture loss concentrates the nutrients, so using the raw ingredient weight as the finished weight understates every nutrient per serving.
Should I lab-test before making a nutrition claim?
If the product’s value is close to the claim’s limit, yes. Test several lots so you know batch-to-batch variation won’t push you over.
Can I use supplier nutrition data?
Yes, and for commercial ingredients it’s often more accurate than a generic database entry.
Does one lab test guarantee my label is accurate?
No. It describes the samples tested. Testing several representative lots gives you a much better picture.
The Bottom Line
You don’t need a lab test to make a Nutrition Facts label. For most formulated foods, a careful recipe calculation is accurate, inexpensive, and easy to update.
Lab testing adds value when processing is unpredictable, ingredients vary a lot, a claim sits near its limit, or a customer requires it. Even then, it works best alongside your recipe calculation, not instead of it.
Calculate first. Test when testing tells you something the recipe can’t.
References
- FDA: Guidance for Industry: Guide for Developing and Using Data Bases for Nutrition Labeling
- 21 CFR 101.9: Nutrition labeling of food (compliance is paragraph (g))
- FDA: Food Labeling Guide
- FDA: Industry Resources on the Changes to the Nutrition Facts Label
- USDA Agricultural Research Service: FoodData Central
- AOAC INTERNATIONAL: Official Methods of Analysis
Disclaimer: This article provides general information about U.S. nutrition labeling and isn’t legal, regulatory, laboratory, or food-science advice. The right method for determining nutrient values depends on your product, ingredients, process, variability, and claims. Food manufacturers are responsible for making sure nutrient declarations comply with FDA requirements. Consult current FDA regulations and guidance and, when appropriate, a qualified food scientist, regulatory professional, or analytical laboratory.
