Last verified 16 Sep 2026 · Sources
Every Nutrition Facts label has to hold up against the same FDA compliance tolerances, but there are two different ways to get the numbers that go on it: calculating them from ingredient and recipe data (database analysis) or physically testing a sample in a laboratory (proximate analysis). Neither method is required by name in the regulation: 21 CFR 101.9 sets tolerances the declared values must meet, not a mandated method for producing them. The practical differences are cost, turnaround, and how directly each method ties your declared values to what’s actually in the package. Checked against the current text of 21 CFR 101.9 and lab-industry sites, September 2026.
Quick answer
Database analysis (pulling nutrient values from USDA FoodData Central and supplier spec sheets, then applying 21 CFR 101.9’s rounding and %DV rules) is the faster, lower-cost route and is what NutriLabels’ $29 Single Product and $249 done-for-you label use. Laboratory proximate analysis physically tests a sample of the finished product and is generally priced per sample by the lab, with most labs quoting rather than publishing a price list. Both methods are allowed under FDA rules: the regulation cares about whether your declared values meet its 80/120-style tolerances (21 CFR 101.9(g)(4)-(5)), not about which method produced them. NutriLabels quotes a 100 g nutrient analysis on request on top of its database-driven build if you want a measured data point without commissioning a full independent lab study.
Side by side
| Method | Typical cost | Turnaround | Allowed under FDA rules? | Best for |
|---|---|---|---|---|
| Database analysis (USDA FDC + spec sheets) | $29 self-serve (NutriLabels Single Product); $249 for the expert-built label | Self-serve is instant; expert tiers ship within 5 business days after documentation is confirmed | Yes. The regulation sets tolerances, not a required method; declared values calculated this way must still meet 21 CFR 101.9(g) | Recipes with stable, known ingredients and available spec sheets; most small packaged-food brands |
| Laboratory proximate analysis | Commercial labs quote proximate analysis per sample after contact; none of the lab sites checked publishes a price list | Lab-dependent; not published on the pages checked this session; request lab-specific turnaround with your quote | Yes. Same 21 CFR 101.9(g) tolerances apply regardless of method | Recipes with novel or unusual ingredients not well represented in a nutrient database, or when a retailer/co-packer specifically requires lab-tested values |
| NutriLabels done-for-you label + 100 g analysis | $249 flat for the label; the 100 g analysis is quoted on request | Within 5 business days after documentation is confirmed, two revisions within 30 days of delivery | Yes | Brands wanting a measured data point layered onto a database-built panel |
Where they differ
What the regulation actually asks for. 21 CFR 101.9(g) doesn’t specify database analysis or lab testing by name. It sets compliance classes: naturally occurring (Class II) nutrients like vitamins, minerals, protein, carbohydrate, and dietary fiber “must be at least equal to 80 percent of the value for that nutrient declared on the label,” while calories, sugars, total fat, saturated fat, cholesterol, and sodium may not run “greater than 20 percent in excess of the value… declared on the label,” both subject to recognized analytical variability (21 CFR 101.9(g)(4)-(5), quoted from law.cornell.edu/cfr/text/21/101.9). Added (Class I) nutrients in fortified foods have to be formulated at least equal to the declared value (21 CFR 101.9(g)(4)(i)). Either method can produce values that meet or fail this test; the regulation checks the outcome, not the process.
Cost transparency. NutriLabels publishes a flat price for every tier: $29 self-serve, $249 for an expert-built label; a 100 g analysis is quoted on request. Commercial labs quote proximate analysis per sample after contact; none of the lab sites checked publishes a price list.
Turnaround. NutriLabels’ self-serve build is instant, and its expert-built tiers ship within 5 business days after documentation is confirmed. Lab turnaround depends on the lab, the specific tests ordered, and current sample volume, and wasn’t published as a fixed number on the sites checked. A lab quote typically includes a turnaround estimate alongside the price.
What each method is actually measuring. Database analysis calculates nutrient values from known ingredient data, useful when your ingredients and their nutrient profiles are well documented (a common spice blend, a standard flour) but less precise for a genuinely novel ingredient or a recipe that varies meaningfully batch to batch. Lab testing physically measures a submitted sample, which ties the declared values directly to that specific batch but doesn’t automatically reflect a future batch made with different lots of ingredients.
Who requires which. Neither the FDA nor NutriLabels’ own methodology dictates that one method is mandatory. Some co-packers, retailers, or specific product categories may ask for lab-verified values as a condition of doing business, but that’s a buyer or retailer requirement, not an FDA one, and worth confirming directly with whoever is asking.
Who each fits
Database analysis tends to suit you if you run: a small or growing food brand with a defined recipe and available ingredient spec sheets, and you want a compliant label quickly and at a known, flat price: the situation NutriLabels’ self-serve and expert tiers are built for.
Lab testing tends to suit you if you have: an unusual ingredient not well represented in nutrient databases, a recipe that needs a measured baseline for your own quality control, or a specific requirement from a co-packer or retailer that values come from an independent lab. Talk to a food-labeling attorney or your co-packer’s compliance team if you’re unsure which one your specific buyer requires.
Choose
A database-built label from a well-documented recipe gives you values you can check against the same 21 CFR 101.9(g) tolerances a lab result faces, without waiting on a lab quote. If you run multiple SKUs or manage labels for wholesale or co-packing clients, the Studio plan ($299 a year: every product you make, plus the compliance binder and PDF reports, AI spec-sheet import, recipe costing, product approval and three user seats) fits, and the Co-packer plan ($1,199 a year: everything in the Studio plan plus client folders, white-label documents and five seats) covers client brands. For a single product, enter your recipe, then see your label: your calories, the claims you qualify for and your allergen line are free, no card required, and Single Product ($29) unlocks every other nutrient value and %DV, every FDA format your product qualifies for, the ingredient statement and allergen declaration. If you want a measured data point added, send us the recipe and we build it: the done-for-you label ($249) builds the panel in every format your product qualifies for, and a 100 g analysis can be quoted on top. See pricing, or build your label free to see the free preview first.
FAQ
Does the FDA require lab testing for a Nutrition Facts label?
No. 21 CFR 101.9 doesn't mandate a specific analysis method. It sets the compliance tolerances a label's declared values must meet, whether they come from database analysis, recipe calculation, or laboratory testing.
What's the 80/120 rule?
It's a shorthand for the compliance tolerances in 21 CFR 101.9(g)(4)-(5): naturally occurring (Class II) nutrients must test at least 80% of the declared label value, while calories, sugars, total fat, saturated fat, cholesterol, and sodium may not test more than 20% over the declared value, subject to recognized analytical variability (law.cornell.edu/cfr/text/21/101.9).
How much does lab testing cost?
Most commercial labs checked (Anresco, and others) do not publish a price list; proximate analysis and full Nutrition Facts panel testing are quoted per sample after a request. Get a direct quote for your product before budgeting.
Is database analysis less accurate than lab testing?
They measure different things: database analysis calculates values from ingredient data (USDA FoodData Central plus supplier spec sheets), while lab testing physically measures a sample. Both are used to meet the same 21 CFR 101.9(g) tolerances; which is appropriate depends on your ingredients, your co-packer's or retailer's requirements, and how much your recipe varies batch to batch.
Can I combine both?
Yes. NutriLabels' done-for-you label ($249) builds the panel using database analysis, and a 100 g nutrient analysis can be added on request (quoted per product), which is one way to add a measured data point to a database-built panel without commissioning a full independent lab study.
More comparisons
Multiple, Multiple, ReciPal, Genesis Foods (Trustwell, formerly ESHA Genesis R&D). See the full all comparisons.