Heavy Metals in Baby Food: What FDA's Lead Action Levels Mean for Manufacturers

A 2021 report from the House Subcommittee on Economic and Consumer Policy tested 168 commercially available baby food products and found detectable levels of heavy metals in 95% of them. That number set off a firestorm — congressional hearings, front-page coverage, class-action lawsuits, and, eventually, formal FDA action. In January 2025, FDA finalized specific action levels for lead in processed foods for babies and young children as part of its multi-year “Closer to Zero” initiative. If you’re making or co-manufacturing infant food in the U.S., you’re now operating under enforceable benchmarks, not suggestions.

Here’s what those limits mean, what the testing actually involves, and where manufacturers typically stumble.

What FDA’s “Closer to Zero” Program Actually Requires

The Closer to Zero initiative, launched in April 2021, targets four elements: lead, inorganic arsenic, cadmium, and mercury. FDA’s approach isn’t to eliminate these contaminants — that’s not technically feasible, since they occur naturally in soil and water — but to drive levels “as low as reasonably achievable” (ALARA) through enforceable action levels.

For raw material and ingredient-level verification, Ayah Labs specializes in contract testing and supplier qualification.

The lead action levels finalized in January 2025 set two tiers:

• 10 ppb (µg/kg, wet weight) for most processed baby foods, including pureed and jarred fruits, vegetables, grain and meat mixtures, yogurts, and custards
• 20 ppb for single-ingredient root vegetables (like pureed sweet potatoes and carrots) and dry infant cereals

These aren’t the same as formal tolerances under 21 CFR Part 109, which establish legal adulteration limits. Action levels operate differently — exceeding one signals that FDA may take regulatory action, but the product isn’t automatically deemed adulterated by statute. That distinction matters for legal strategy, but in practice, if your finished product tests above 10 ppb for lead, you have a compliance problem.

For inorganic arsenic, FDA already finalized an action level of 100 ppb for infant rice cereal back in 2020 — and that one has teeth. Several manufacturers have already received warning letters tied to arsenic exceedances in rice-based products. Cadmium and mercury action levels for infant foods are still in development under the Closer to Zero roadmap.

Why Baby Food Matrices Are Analytically Harder Than They Look

Anyone who’s run a metals panel on infant food knows this: the matrices are a pain. Pureed sweet potato doesn’t behave the same as apple puree, which doesn’t behave the same as a rice-oat cereal blend, which certainly doesn’t behave the same as a meat-and-vegetable puree with 8% fat. Each matrix affects acid digestion efficiency, and poor digestion means poor metal recovery — which means you can test a product that’s actually noncompliant and get a passing result.

The standard approach for trace metals in food is ICP-MS (inductively coupled plasma-mass spectrometry) following microwave-assisted acid digestion. ICP-MS detection limits for lead in food matrices typically run between 0.5 and 2 ppb under well-optimized conditions, which gives you the sensitivity headroom you need to confirm compliance at the 10 ppb threshold. But instrument capability is only half the story.

Recovery validation is where things get serious. A credible metals testing program runs certified reference materials (CRMs) with every analytical batch — materials like NIST SRM 1566b (oyster tissue) or NIST SRM 2383 (baby food composite) that have independently assigned values for each target element. If your laboratory isn’t hitting 85–115% recovery on those CRMs, the numbers coming out of that instrument aren’t reliable, regardless of the detection limit on the spec sheet.

At Qalitex, we run ICP-MS metals panels with NIST-traceable CRMs as a standard part of every infant food submission. We’ve seen products from suppliers with in-house QC programs produce wildly inconsistent recoveries because their digestion protocols weren’t validated against the specific matrices they actually produce. It’s not a calibration issue — it’s a chemistry issue, and it’s fixable, but only if you know to look for it.

Where Manufacturers Actually Get Caught

The contamination story in infant food usually isn’t about process contamination — it’s about raw ingredients. Root vegetables bioaccumulate lead and cadmium from soil. Rice absorbs inorganic arsenic through its root system far more efficiently than other grains. Fruits grown near legacy industrial sites or old orchards can carry elevated lead loads even decades after agricultural pesticide use stopped.

That means finished product testing alone is a rearguard action. By the time a batch of sweet potato puree comes back at 14 ppb lead, you’ve already committed to a full production run, and the corrective path is usually destroy-and-remanufacture. The smarter model is raw material qualification testing — running metals panels on incoming ingredient lots before they enter production.

This is where I see the biggest gap in smaller manufacturers’ programs. A Subpart B supplier qualification framework under 21 CFR Part 117 technically requires hazard analysis for contaminants like heavy metals, but many operations rely entirely on certificates of analysis (COAs) from their ingredient suppliers. COAs are not equivalent to independent testing. A COA from a broker may be months old, may reflect a different crop lot, or may use a method without adequate sensitivity for a 10 ppb threshold. It is, essentially, paper compliance.

If your hazard analysis identifies lead, arsenic, or cadmium as hazards requiring a preventive control, independent incoming lot testing should be part of that control — not just a review of whatever the supplier sends you.

The Cadmium and Mercury Gap Nobody Is Talking About Enough

Lead and arsenic get most of the regulatory attention because FDA has finalized action levels for them. But the cadmium picture in infant food is quietly concerning, and manufacturers who are only running two-element panels right now are building a compliance gap they may not see coming.

The 2021 House Subcommittee report found cadmium in 73% of the tested products. Cadmium is a known nephrotoxin and probable human carcinogen (IARC Group 2A), and the dietary accumulation concern for infants is real — young children absorb cadmium from food at roughly five times the rate adults do, according to EFSA risk assessments. FDA has not yet finalized cadmium action levels for infant food, but the Closer to Zero roadmap explicitly includes it as a priority element. The manufacturers doing four-element panels now won’t be scrambling when those limits land.

Mercury deserves mention too, particularly for products containing fish-derived ingredients. Salmon, tuna, and anchovy components show up in pouches and purees marketed to toddlers with some regularity. FDA’s Closer to Zero analysis identified mercury as a priority element across the infant food category. Running a total mercury screen on any product with marine ingredients is a reasonable baseline regardless of whether FDA has finalized a specific action level — because the data you’re generating now will be the baseline you need to demonstrate compliance later.

A 4-element panel (lead, inorganic arsenic, cadmium, and total mercury) on representative finished products at least quarterly, and on incoming raw material lots with elevated risk profiles, is defensible protocol for 2026.

What Adequate Documentation Looks Like Under Scrutiny

FDA’s enforcement posture under Closer to Zero involves records review as much as field sampling. When investigators come in, they’re going to ask for your hazard analysis, your supplier qualification records, and your finished product testing history — and they’re going to want to see how your testing results connect to your corrective action procedures.

“We test every batch” isn’t enough if the results live in a spreadsheet no one reviews, with no out-of-specification thresholds defined and no documented response protocol. A metals testing program that will hold up under inspection should include:

1. Written specifications for each target element, for each product category, tied explicitly to the applicable FDA action level
2. Test records with traceability — instrument run data, CRM recovery data, analyst sign-off, and lot identification linking back to the production record
3. A documented OOS procedure that specifies when a lot is quarantined, when a root cause investigation is triggered, and what the disposition criteria are before a lot can be released or destroyed
4. Supplier qualification trend data showing how raw material testing results are tracked over time, not just reviewed batch by batch in isolation

An ISO 17025-accredited laboratory can provide test reports that satisfy chain-of-custody and defensibility requirements for regulatory submissions. That matters if FDA’s findings ever become the basis of a recall, a warning letter, or civil litigation — all of which have happened in this category in the past five years.

---

The 10 ppb lead threshold is tighter than anything the infant food industry has operated under before. A meaningful portion of products that would have cleared any previous benchmark won’t clear this one. The manufacturers who find that out through their own proactive testing — before an FDA field sample or a third-party investigative report — are in a fundamentally different position than those who find out from a warning letter. Start with your highest-risk raw materials: root vegetables, rice ingredients, and any imported commodity with limited origin traceability. Run independent lot testing, trend the results, and document the program. That’s not overcompliance. That’s just the current standard for selling food to infants.