FDA’s fiscal-year 2018–2020 Total Diet Study analytical-results file contains 68,993 results from 3,276 food, beverage and water composites across 307 TDS foods. The data preserve each food’s preparation, collection, analyte, original unit, concentration and reporting limit. The complete analytical file and FDA’s independent summary workbook have been reconciled at food × analyte level; FDA’s narrative report describes the study design. Total Diet Study Report: Fiscal Years 2018-2020 Elements Data
Key numbers
| Measure | Verified scope |
|---|---|
| Analytical results | 68,993: 68,831 element results and 162 arsenic-speciation results |
| Composite samples | 3,276, including 3,241 food/beverage composites and 35 bottled-water composites |
| TDS foods | 307, with one explicit ingredient-destination route per food |
| Analytes | 28 across the complete file; individual panels differ between foods, water and the selected speciation subset |
| Inorganic arsenic | 54 results across 10 food identities; the same subset also has dimethylarsinic acid and monomethylarsonic acid results |
| Food × analyte summaries | 6,479: 6,449 independently reconciled against FDA’s published workbook and 30 speciation groups calculated from the analytical records |
| Reporting periods | Fiscal years 2018–2020; calendar years 2017–2020 |
The examples below illustrate the dataset’s different matrices and sample sizes. These are source-specific observations, not estimates of the full retail distribution. Concentrations are in µg/kg (ppb) on the food form named in the first column.
| TDS food and preparation | Analyte | Composites analysed | Detects | Source result |
|---|---|---|---|---|
| Whole fluid milk, food 1 | Pb | 27 | 0 | All non-detects; reporting limit 1 ppb; FDA reports mean and SD as N/A |
| Chocolate-chip cookies, food 183 | Pb | 3 | 3 | Detected range 5.6–10 ppb; FDA-reported mean 7.6 ppb |
| Infant rice cereal, dry, food 765 | iAs | 9 | 9 | Detected range 62.2–85.8 ppb; calculated source-sample mean 72.8 ppb |
| Crisped rice cereal, food 75 | iAs | 3 | 3 | Analytical records 85.8, 90.4 and 103 ppb; calculated mean 93.0667 ppb |
| Canned tuna in water, drained solids, food 340 | tHg | 3 | 3 | Detected range 220–250 ppb; FDA-reported mean 230 ppb |
| Cocoa powder, food 910 | Cd | 1 | 1 | Single result 400 ppb; FDA reports mean and SD as N/A |
| Baking powder, food 902 | Pb | 1 | 1 | Single result 164 ppb; FDA reports mean and SD as N/A |
The narrative report’s crisped-rice-cereal range begins at 8.9 ppb, which disagrees with the three primary analytical records. This page uses the analytical records and preserves the discrepancy as a source-context disposition; no source file has been altered. Total Diet Study Report: Fiscal Years 2018-2020 Elements Data
Methods (brief)
FDA sampled regional and national food lists, prepared foods in their stated form and analysed composites. Regional collections represent three cities in a region; national collections use one collection point. Fiscal-year 2019 collections were interrupted by the federal government shutdown and fiscal-year 2020 collections by the COVID-19 pandemic. Foods are kept separate by FDA food number, so dry infant cereal, cereal prepared with water and ready-to-eat cereal remain different matrices. Total Diet Study Report: Fiscal Years 2018-2020 Elements Data
The analytical key defines fiscal year, calendar year, collection, food number and description, season, food-list type, region, analyte, units, concentration and reporting limit. FDA’s reporting limits are administrative limits based on LOD and LOQ across analyses. Original values and units are retained; ppm values are also expressed as ppb using the exact factor 1,000. No moisture or reconstitution conversion is applied. FY2018-FY2020 TDS Elements Analytical Results Key
Reported zero codes are non-detects, not measurements of zero concentration. Each result carries its reporting limit and detected/non-detect status. FDA’s narrative report states that non-detects were included as zero when calculating means; the summary workbook retains explicit ND and N/A entries. Positive reported measurements are retained even if below the administrative reporting limit printed for the row, and detection counts are checked against FDA’s independently published counts. Total Diet Study Report: Fiscal Years 2018-2020 Elements Data
The downloadable summary labels calculated lower-bound statistics using ND = 0 and supplies corresponding upper-bound estimates using each non-detect’s reporting limit. Its descriptive percentiles use linear interpolation at position (N − 1) × p. These are source-scope estimates; an all-non-detect group does not establish that its true median, upper percentile or maximum is zero. FDA-reported means and SDs are kept separately from calculated values, including N/A where FDA did not report a statistic.
Evidence fitness
The dataset supports traceable occurrence findings for the particular food, analyte, preparation and collection period tested. All 6,449 regular-element summary groups agree with FDA’s workbook on sample count, detect/non-detect count and reported extrema. Source means and SDs also reconcile within FDA’s stated rounding. The 30 arsenic-speciation summaries are calculated from the original records because they are not tabulated in that supplement.
It does not identify brands or lots, establish current brand performance, or support substituting an ingredient concentration for a mixed-food concentration. Some foods have only one to three composites; source-specific ranges and means do not establish a stable market-wide tail. Compositing, censoring and selection for arsenic speciation must be retained when interpreting the findings. This dataset does not itself establish product compliance or certification thresholds.
Limitations
Chromium is total chromium, not Cr(VI). Mercury is total mercury, not methylmercury. Total arsenic and inorganic arsenic remain distinct. Inorganic arsenic was measured only for selected rice-containing foods, grape juices and white wine; it cannot be attributed to foods with only total-arsenic measurements. The numerical basis remains the stated tested form: powdered infant formula and formula prepared with water are separate food identities, as are dry and prepared infant cereals.
Structured data and routing
Download all analytical results, food × analyte summaries, food-to-ingredient routes and the source reconciliation. The downloads contain the source identity, exact food matrix, units, censoring, statistic type and source-row or summary-workbook locator needed to reproduce a finding.
Every one of the 307 food identities has an explicit destination route. A destination may collect several food preparations as separately labelled observations; their results are not pooled merely because they appear on the same page. Mixed baby foods remain mixed-food observations and are not decomposed into their component ingredients. Product-specific use requires matching the species and tested form to that product.
Verification and correction history
The September 2026 audit found that the earlier 29,148-row extract was an exact prefix of FDA’s complete 68,993-row file and ended partway through the analytical panel for food 183. It represented 90 of 307 foods. The earlier description of the supplement as already fully represented by that extract was incorrect.
The complete original FDA analytical file and summary workbook were retrieved on 2026-09-10, preserved with SHA-256 hashes, and reconciled before replacing the derived extract. This recovered 39,845 analytical results and the 217 previously absent food identities, and completed the partly extracted chocolate-chip-cookie panel. The normalized files now carry explicit non-detect semantics and preserve source-reported mean, SD, detection count and N/A values. Obsolete narrative-derived candidate rows were retained in an audit and superseded by the complete primary extraction.
The June 2026 supplement PDF and its byte-identical duplicate copies remain recorded in the source metadata. Their prior presence did not establish extraction completeness.
Related evidence
Ingredients, Total Diet Study Report: Fiscal Years 2018-2020 Elements Data and FY2018-FY2020 TDS Elements Analytical Results Key. Individual food destinations are listed in the downloadable routing table.
Update history
No substantive edit history is available in this build. The full commit record is available in git.