The FDA reported that the U.S. supply of infant formula is safe following the “most rigorous testing” of the product ever conducted.
The testing program did not include foodborne pathogens such as Cronobacter, Listeria, Salmonella and E. coli. The agency tested for arsenic, cadmium, lead, mercury, 30 PFAS, 318 pesticides, 21 phthalates and one non-phthalate plasticizer.
The investigation was part of Operation Stork Speed, launched in 2025 by the Food and Drug Administration (FDA) to ensure the safety of infant formula. This measure was in response to the 2022 Cronobacter outbreak.
“We are testing more infant formula than ever before, and the results are clear: While most products meet high safety standards, even the smallest exposures matter to newborns,” said HHS Secretary Robert F. Kennedy Jr.
According to the FDA report, the tests “confirmed that the overwhelming majority of samples contained undetectable or very low levels of contaminants, confirming that the U.S. infant formula supply is safe.”
However, in the United States, there are no legal limits on metals in infant formula tested by the FDA. As a result, when the FDA calls a formula “safe,” there is no standard behind those words.
Operation Stork Speed is also investigating the supply of infant formula in the United States, which has been significantly depleted following a recall by Abbott Nutrition due to a Cronobacter outbreak.
“We are committed to ensuring that infants and toddlers have safe, high-quality formula options supported by a resilient supply chain,” said FDA Commissioner Marty Makary.
The FDA tested 312 samples from 16 brands, but the report did not say which brands were tested. The formula types tested were: powdered with 278 samples; 11 concentrates; 23 is ready to dispense liquid. Protein sources sampled were 258 milk, 44 soy milk, and 10 amino acids.
“Across the products tested, most infant formula samples contained undetectable or very low levels of contaminants, confirming the safety of the U.S. infant formula supply,” the FDA report said.
“Although contaminant levels were low, some samples had values that initiated additional follow-up as part of the agency’s robust monitoring and oversight efforts, which will include conducting additional testing and continuing to work with manufacturers on actions to reduce contaminant levels as low as possible.”
For each contaminant reported by the FDA, the following values represent the typical levels measured.
- Not Detected: If present, the amount is too small to be detected by testing methods.
- Range: from the lowest level detected to the highest level (concentration) detected.
- 95th percentile (reported as 95%): The level at which 95% of the sample falls below (meaning only 5% of the sample were higher)
- Median: The midpoint of a set of data when the numbers are ordered from lowest to highest. It helps display common numbers within a group.
- A ppb (parts per billion) is equivalent to one drop of water in an Olympic-size swimming pool.
The test results were as follows:
- mercury It was not detected in 296 (95%) of 312 samples. Concentrations among all samples ranged from undetectable to 0.3 ppb. Due to the high number of non-detections, the 95th percentile and median were not calculated.
- cadmium It was not detected in 106 of 312 samples (34%). Concentrations among all samples ranged from undetectable to 1.5 ppb. 95% of all samples had cadmium levels below 1.1 ppb. The median cadmium concentration was 0.2ppb.
- head It was not detected in 61 of 312 samples (20%). Concentrations among all samples ranged from undetectable to 1.1 ppb, with 95% of samples containing less than 0.5 ppb lead. The median lead concentration was 0.2 ppb.
- arsenic It was not detected in 18 of 312 samples (6%). Concentrations among all samples ranged from undetectable to 4.7 ppb. 95% of all samples contained less than 2.0 ppb arsenic. The average arsenic concentration was 0.5 ppb.
- Pesticides: Each sample was tested for 318 different pesticides, including glyphosate and glufosinate. No pesticides were detected in 309 samples (99%), and glyphosate and glufosinate were not detected in any samples. Due to the high number of non-detections, the 95th percentile and median were not calculated. Only three samples had detectable levels of pesticides, all at very low concentrations. One sample had 0.25 ppb of chlorpyrifos, 0.20 ppb of malathion in another, and 0.20 ppb of piperonyl butoxide in a third.
- PFAS: Each sample was tested for 30 different PFAS compounds. Most PFAS compounds (25 out of 30) were not found in any of the samples. Five PFAS were detected in a subset of samples at concentrations ranging from 0.51 ppt to 150 ppt, depending on the compound. The most commonly detected one was perfluorooctanesulfonic acid (PFOS), which was found in half of the samples at detectable concentrations ranging from 0.51 to 6.0 ppt, with 95% of all samples containing less than 2.9 ppt of PFOS. Due to the large number of non-detections, the median was not calculated. PFAS levels are often reported in parts per trillion (ppt) because testing methods can detect these contaminants even at extremely low concentrations. One trillionth of a trillionth is equivalent to one second in 32,000 years, or one drop in 1,000 Olympic-size swimming pools.
- Phthalates: Each sample was tested for 21 phthalates and 1 non-phthalate plasticizer. Phthalates were not detected in 167 of 312 samples (54%). Among the compounds tested. None of the seven phthalates were detected in any of the samples. Five phthalates were barely detected (less than 2% of samples). Concentrations of the remaining compounds detected ranged from undetectable to 145 ppb. The most commonly detected phthalate compounds were DEHP and DINP.
- DEHP It was not detected in 253 of 312 samples (81%). DEHP concentrations in the samples ranged from undetectable to 57 ppb. Due to the high number of non-detections, the 95th percentile and median were not calculated. DINP was not detected in 260 (83%) of 312 samples. DINP concentrations in the samples ranged from undetectable to 145 ppb. Due to the high number of non-detections, the 95th percentile and median were not calculated.
- Breast milk: Although this testing phase focused specifically on infant formula products, the FDA also tested a limited number (110) of breast milk samples from a single state donor milk bank for arsenic, cadmium, mercury, and lead. Overall, the majority of samples (85%) had at least one detectable contaminant, while 17 samples (15%) had no detectable contaminants. The detection rate was highest for mercury at 61%, followed by cadmium at 57%, arsenic at 33%, and lead at 30%.
