baby food

worst-offender brands, ranked by average lead

top 20 brands across baby food by average lead concentration in ppb. minimum 5 lead measurements per brand to qualify (so a single bad lot can't dominate). this is cross-source, AB 899 lab disclosures, HBBF independent testing, NYC inspector samples, all combined. 0 ppb readings count too: a brand that tests at the limit of detection on every lot will rank low here, which is the point.

worst-offender lots

top 20 individual lots by lead concentration, all sources combined. each row is one production batch tested in one lab. all four heavy metals shown side-by-side so you can see whether lead is the only problem or whether arsenic, cadmium, and mercury are tracking with it. for AB 899 rows the lot code is the manufacturer's own batch identifier, that's the number you'd find on the bottom of the jar.

all values in ppb (µg/kg). dashes mean the source did not report that metal for that lot. Pb = lead, As = arsenic, Cd = cadmium, Hg = mercury.

per-ingredient breakdown

this is the cleanest signal in the entire dataset. lead doesn't track with brand, it tracks with ingredient. fig is the worst at 52.7 ppb avg lead, driven by one brand's oat-bar line. sunflower seed, almond, and wheat follow. on the staple end, sweet potato (1,963 measurements, 18 brands) and spinach (1,310 measurements, 14 brands) cluster around 3.5 ppb, low individually, but they're in almost every product so they drive a lot of total exposure.

derived from the AB 899 dataset (~18,124 lots tested for lead). ingredient is parsed from product name. this is the same analysis layer as /pages/babyfood but rendered here so you can sort it next to the cross-source brand and lot views.

by product format

same data sliced by product type instead of ingredient. bars, nut butters, and snack crunches top the list (because of the nut, seed, and rice content). traditional pouch purees and cereals trend lower per measurement, but each kid eats more of them so total exposure can still add up.

sources contributing to baby food

nearly all of the data on this page comes from california's AB 899 disclosure law. it's the largest single contribution to the entire lead database, 47,802 measurements across 828 products from 31 brands, refreshed every quarter. the rest is independent testing (HBBF) and a handful of NYC inspector samples, FDA recalls, and one king-county and one pure earth measurement.

• AB 899
  47,802 measurements
  california state law forcing every baby food maker selling in CA to publish lab results for arsenic, cadmium, lead, and mercury for every lot. effective jan 2025. 47,802 measurements across 828 products from 31 brands.
  read about AB 899 →
• HBBF 2019
  456 measurements
  168 baby foods purchased off store shelves and independently tested for arsenic, cadmium, lead, and mercury. the report that changed the public conversation about heavy metals in baby food. 456 measurements.
  read about HBBF 2019 →
• NYC NPSD
  17 measurements
  NYC inspectors have been buying spices, cosmetics, supplements, cookware, religious products, candies, toys, and anything else they find suspicious off bodega and import-store shelves since 2008, then lab-testing them. 8,494 measurements across 12 categories from 121 countries.
  read about NYC NPSD →
• FDA
  2 measurements
  federal recall notices for food, drug, and cosmetic products. also reactive, products pulled after a problem was identified. mostly spices, supplements, and a slice of baby food. 218 records.
  read about FDA →
• King County
  1 measurements
  puget sound county hazardous waste program. store-shelf surveys for cookware, dishes, cosmetics, spices, jewelry, and toys. 2,197 measurements covering 1,580 products from 668 brands and 61 countries, 2019 to 2025.
  read about King County →
• Pure Earth
  1 measurements
  global NGO (formerly blacksmith institute). market surveys for cookware, spices, cosmetics, paint, and toys across 27 countries, mostly low- and middle-income markets where lead enforcement is weak. 5,153 measurements covering 3,472 products.
  read about Pure Earth →

two layers, one dataset

researcher view: the raw table

the full baby food dataset, all 48,279 rows, sortable, filterable, paginated. one row per (lot, metal). search by brand or product, filter by metal / source / country. units are ppb (parts per billion). 1,000 ppb = 1 ppm = 1 mg/kg = 0.0001%. so a value of 10,000,000 ppb would be 1% lead by mass (none of the baby food rows reach anywhere near that, the worst lot here is 35,470 ppb, a single salad-bowl entry from pure earth's india work).