Dr. Manouchehr Hessabi
← All writing
7 min readenvironment · pediatric · methods

What an FDA action level for metals in baby food means

FDA proposed cadmium action levels for baby food in October 2026: what an action level is, how FDA sets one, and what it does not say about safety.

By Manouchehr Hessabi, MD, MPH

On October 23, 2026, the U.S. Food and Drug Administration (FDA) published draft action levels for cadmium in processed foods for babies and young children. They join the action levels for lead that FDA finalized in January 2025. It is easy to read numbers like these as safety lines, a point below which a food is fine and above which it is dangerous.

That reading is not quite right. The difference is a useful lesson in how exposure science becomes regulation, and it is the same distinction that matters in research on environmental exposures and child development: what a number was built to do determines what it can tell you.

What is an FDA action level?

An action level is a concentration of a contaminant at which FDA may regard a food as adulterated, meaning legally unfit under the Federal Food, Drug, and Cosmetic Act. It is not a binding limit. FDA says it will weigh an action level "in addition to other factors" when deciding whether to take enforcement action in a particular case, and the agency states plainly that it "does not need an action level for a heavy metal to take enforcement action."

The legal basis is a regulation, 21 CFR 109.6, and the levels themselves are issued as guidance. In the words of FDA's final guidance on lead, its guidance documents "do not establish legally enforceable responsibilities."

The levels are expressed in parts per billion (ppb). One ppb is one part of a substance in a billion parts of food by weight, equivalent to one microgram of metal per kilogram of food.

What did FDA propose for cadmium in October 2026?

The draft guidance, announced in the Federal Register on October 23, 2026 (Docket FDA-2025-D-4023), covers processed food intended for babies and young children less than 2 years old. The proposed levels are:

  • 10 ppb for fruits, vegetables (excluding single-ingredient carrots), and mixtures, including grain- and meat-based mixtures
  • 25 ppb for single-ingredient carrots
  • 45 ppb for dry infant cereals
  • 20 ppb for yogurts, puddings, and custards
  • 25 ppb for single-ingredient meats

These are proposals, not final values. The notice asks for comments by December 22, 2026, including on "whether there may be sufficient data or information that would support different levels for these categories of foods." The final numbers may differ.

How do the cadmium levels compare with the final lead levels?

FDA's January 2025 lead guidance set two tiers:

  • 10 ppb for fruits, vegetables (excluding single-ingredient root vegetables), mixtures, yogurts, custards and puddings, and single-ingredient meats
  • 20 ppb for single-ingredient root vegetables and for dry infant cereals

The structures are similar. Both metals have a general level of 10 ppb for most categories and higher levels for a small number of specific categories, with dry infant cereals and a root vegetable category among them in both cases.

It is tempting to read a higher number as a judgment that a food is more dangerous, or a lower one as a statement that it is safer. Neither inference follows from the number alone. As the next section explains, FDA's stated basis for an action level includes what manufacturers can achieve, and that input can differ from one food category to the next for reasons unrelated to how risky the food is to eat.

How does FDA arrive at a number?

FDA's Closer to Zero page (content current as of October 27, 2026) says action levels are based on "toxicological data and what is achievable by manufacturers." Those are two different kinds of evidence, and it helps to look at each. The toxicological side rests on dose-response reasoning, explained in the piece on dose-response relationships.

The toxicological benchmark. This is a reference point for dietary intake derived from health research.

  • For lead, FDA uses an interim reference level (IRL), which the agency calculated from the blood lead reference value of the Centers for Disease Control and Prevention (CDC). FDA describes the CDC value as "the level at which the CDC recommends clinical monitoring of lead exposure in children." The derivation is set out in peer-reviewed papers that FDA lists, including the original IRL analysis and an updated version in Regulatory Toxicology and Pharmacology.
  • For cadmium, FDA derived a toxicological reference value (TRV), which describes the relationship between dietary cadmium exposure and adverse health effects. The method is published as a reassessment of the cadmium TRV in the same journal.

Achievability. The second input is what manufacturers can reach in practice. The Federal Register notice describes the cadmium draft as "intended to set achievable action levels that will help further reduce cadmium in the food supply."

The method point is worth stating directly. A number set partly by achievability describes the food supply as much as it describes biology. It is a target that pulls the upper end of the distribution down. It is not a threshold below which exposure has been shown to be harmless, and it was not designed to be one.

This is also why the regulatory number and the health evidence belong in separate discussions. Readers who want the health side can find it in the explainers on what the evidence shows on cadmium exposure in children and on lead exposure in children.

Why does presence not equal risk?

FDA's own wording is direct: "Just because a heavy metal is present doesn't mean a product is unsafe."

Epidemiologists distinguish the presence of a substance from exposure, and exposure from dose. What matters for health is how much reaches the body, how often, over what period, and alongside what else in the diet. A single measurement of one food is a starting point for that question, not the answer.

Action levels work at the level of the food supply. They are meant to lower average exposure across many children over time, which is a population-health goal rather than a verdict on any one jar.

Why did FDA not set a level for mercury?

The Closer to Zero page explains that FDA tested 995 samples of foods for babies and young children and found that "only 4 samples had total mercury levels above the level of quantitation," the lowest concentration a laboratory method can measure reliably. On that basis, FDA says it is "not prioritizing development of action levels for mercury" in these foods for now.

This is a data-driven choice, and it illustrates that the program is not a blanket rule applied to every metal. Mercury exposure in childhood is a separate question, covered in the explainer on mercury in fish and child development.

What remains uncertain?

Several things are not settled, and an honest reading keeps them in view.

  • The cadmium levels are a draft. FDA has asked whether different levels could be supported, and the final guidance may change.
  • Arsenic is still to come. FDA says it is targeting 2027 to release draft guidance on inorganic arsenic in processed baby food.
  • The scope is processed food for children under 2. The levels apply to that category; they do not describe every food a young child eats.
  • Monitoring has limits. Sampling captures a set of products at a point in time, so results describe what was tested, not every product on a shelf.
  • The program is a cycle. FDA describes four stages (evaluate the science, propose levels, consult stakeholders, finalize) and says it will continue "this cycle of review, consultation, and improvement as needed."

A measured close

An action level is a regulatory tool with a specific job: it gives FDA a reference point for enforcement and gives manufacturers a target to meet. It combines a health-based benchmark with what industry can achieve, which is why it should be read as a statement about the food supply, not as a line between safe and unsafe.

That distinction, between what a number measures and what it is used to claim, runs through good environmental epidemiology. Readers interested in how exposure is measured and interpreted in child-health research can find related peer-reviewed work on the publications page.

About the author. Dr. Manouchehr Hessabi is a physician-epidemiologist and Senior Research Scientist at the BERD core of UTHealth Houston's Center for Clinical and Translational Sciences. See his peer-reviewed publications or research programs.