FDA CFSAN 2023 — Action Level for Inorganic Arsenic in Apple Juice: Guidance for Industry

FDA’s June 2023 final guidance establishes an action level of 10 µg/kg (10 ppb) for inorganic arsenic in single-strength apple juice intended for human consumption, including apple cider. This finalises the approach presented in the July 2013 draft guidance (docket FDA-2012-D-0322) and is an enforcement reference under 21 CFR 109.6(d): apple juice with iAs at or above this level may be regarded as adulterated under section 402(a)(1) of the FD&C Act (21 U.S.C. 342(a)(1)). The action level was set on the basis of achievability under current good manufacturing practices together with a quantitative risk assessment that considered both cancer (lung and bladder) and non-cancer (neurodevelopmental) endpoints.

Key numbers

Action level: 10 µg/kg (10 ppb) iAs in single-strength apple juice (and apple cider). Regulatory basis: 21 CFR 109.6(d); adulteration provision 21 U.S.C. 342(a)(1); docket FDA-2012-D-0322. Historical reference: FDA’s 2008 hazard assessment used a 23 ppb iAs level of concern based on short-term, non-cancer endpoints; that level of concern is no longer in use (superseded by this 2023 chronic-exposure action level).

FDA surveillance data cited in the guidance (Tables 1-2):

• FY2011 retail sample survey (n=94, all speciated): mean iAs 4.4 ppb (SD 2.4), range 0-9.8 ppb. Mean total As 6.4 ppb (SD 4.7), range 1.3-36 ppb. 90 of 94 samples had total As ≤10 ppb; the remaining four had total As 11-36 ppb but all four had iAs <10 ppb. All 94 samples had iAs <10 ppb; 51 samples had iAs <5 ppb.
• FY2013-FY2022 TEP data (n=160 single-strength apple juice samples; total As): mean 4.6 ppb (SD 6.6), range 0-44 ppb. 143 samples had total As ≤10 ppb; 125 samples had total As ≤5 ppb.
• FY2013-FY2022 TEP speciation subset (n=14, drawn from the 160 above): mean iAs 20.6 ppb (SD 10.1), range 8.3-39 ppb; 12 of the 14 confirmed iAs >10 ppb. (Speciation was applied selectively to samples with elevated total As, so this subset is not representative of the broader market.)
• FY2013-FY2020 Total Diet Study (n=47 composite samples, including babies-and-general-consumption): mean total As 2.7 ppb (SD 2.0), range 0-10 ppb. TDS samples are three-retail composites and were not used in the achievability assessment.

Achievability assessment (Tables 3-4; percentage of samples at or below each hypothetical limit):

• FY2011 dataset (inorganic arsenic basis): 31% at 3 ppb, 54% at 5 ppb, 100% at 10 ppb.
• FY2013-FY2022 dataset (total arsenic basis as conservative proxy): 63% at 3 ppb, 78% at 5 ppb, 89% at 10 ppb.

Modeled lifetime (0-50 year chronic) lung+bladder cancer disease rates from Table 3 (FY2011 iAs-basis, cases per million):

• 3 ppb limit: 2.5 (average consumer), 7.7 (3x high consumer).
• 5 ppb limit: 4.8 (average), 14.9 (high).
• 10 ppb limit: 8.0 (average), 24.9 (high).

Table 4 (FY2013-2022 total-As basis, cases per million):

• 3 ppb: 2.5 (average), 7.5 (high). 5 ppb: 3.2 (average), 9.8 (high). 10 ppb: 4.2 (average), 12.8 (high).

Exposure reduction: a 10 ppb action level reduces estimated dietary total-arsenic exposure from apple juice by 42% at the 90th percentile children’s consumption level (FY2013-2022 dataset).

Codex status: no Codex Alimentarius maximum level for iAs in apple juice at the time of publication.

Methods (brief)

Guidance document compiled from FDA Toxic Elements in Food and Foodware, and Radionuclides in Food (TEP) program surveillance and the FDA Total Diet Study. Analytical methods cited:

• Total arsenic: inductively coupled plasma mass spectrometry (ICP-MS), FDA in-house method (link cited in guidance footnote 9).
• Arsenic speciation: high-performance liquid chromatography coupled to ICP-MS (HPLC-ICP-MS), FDA in-house method (link cited in footnote 9). Speciation distinguishes iAs from organic forms monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA). FDA’s stated practice is to screen for total arsenic and speciate only samples with elevated totals.

Limit of quantitation: 2.8 ppb per arsenic species in the FY2011 work; values below the LOQ were set to 1.4 ppb when computing the FY2011 inorganic-arsenic average. In the FY2013-2022 total-As achievability analysis, non-quantified values below LOQ were imputed at 2 or 4 ppb (the relevant LOQ) and values below LOD were set to zero — a conservative approach that assumes all measured arsenic is inorganic when computing disease rates.

Consumption inputs: NHANES 2007-2008 per-capita apple juice consumption (0.83 g/kg bw/day for all persons aged 0-50) for the FY2011 modeling; WWEIA / NHANES 2003-2018 (0.69 g/kg bw/day) for the FY2013-2022 modeling.

Dose-response: FDA used Taiwanese drinking-water arsenic cancer epidemiology (Chen et al. 2010 lung and urinary tract cancer studies, Refs. 20-21) as the basis for the cancer dose-response model.

Limitations

The guidance is non-binding: FDA states it will consider analytical confidence and other factors when deciding whether to bring enforcement action in a particular case, rather than treating 10 ppb as a strict tolerance. The cancer dose-response is extrapolated from high-exposure Taiwanese drinking-water epidemiology, which is the same evidence base challenged in the NRC 2013 interim report (Ref. 5). The FY2013-2022 disease-rate modeling assumes all measured total arsenic is inorganic — a deliberately conservative substitution acknowledged in the footnote to Table 4 (median iAs/tAs ratio in the FY2011 sample was approximately 0.75). Codex has not set a parallel maximum level for iAs in apple juice, leaving the US standard internationally isolated. Apple juice concentrate is not separately addressed; dilution water quality (≤10 ppb As under 40 CFR 141.62 for municipal supply) is identified as a source of variation.

Implications

• Certification: This is the operative US regulatory ceiling for iAs in apple juice and apple cider. The 10 ppb action level is the regulatory cap input for any HMT&C apple-juice or apple-juice-concentrate row. Manufacturer-side achievability percentages from Tables 3-4 (FY2011 iAs basis: 31% / 54% / 100% at 3 / 5 / 10 ppb; FY2013-2022 tAs basis: 63% / 78% / 89% at 3 / 5 / 10 ppb) are available as occurrence inputs to the Standards Workbench; threshold selection itself is a workbench decision per Part 2 / Part 19, not a wiki claim.
• Courses: Anchor regulatory document for the US iAs-in-juice standard. Useful exemplar of FDA’s use of achievability as a co-criterion with risk; the rationale chain (FY2011 surveillance → 2013 draft → 2023 final) is teachable as a Closer to Zero workflow.
• App: Apple juice as an ingredient carries a 10 ppb iAs regulatory ceiling in the US, with market mean iAs around 4 ppb in routine surveillance. Apple cider falls under the same action level. Concentrate and ready-to-drink reconstituted juice are both in scope.

Verification notes

Merge-enhanced 2026-05-19 from existing 2026-05-12 page during MFK ingest pass (MFK_arsenic-in-pear-juice-analytical-results-from-2005). The PDF filename in the Manual Fetch Kimi folder (“Arsenic in Pear Juice Analytical Results from 2005-2011.pdf”) is mislabeled; the actual content is FDA’s June 2023 Apple Juice action level guidance. The MFK PDF is byte-identical (SHA256 0c92d0eb07e37fe4eecd73904bfcd6a4ceedadd9a90494012820d27b49811454) to the previously-ingested raw/reports/fda2023-ias-apple-juice-guidance.pdf, so the existing cite-key and raw_path are preserved.

Enhancements applied:

• Added raw_handle to record the MFK manual-fetch identity.
• Renamed legacy sha256: field to current raw_sha256: schema.
• Added audience frontmatter.
• Added tAs to metals (the guidance reports total arsenic surveillance data alongside speciated iAs).
• Dropped products/fresh-fruit from products: the guidance is exclusively apple juice and apple cider; fresh fruit is out of scope.
• Removed legacy ## Wiki pages updated on ingest heading.
• Expanded Key numbers with the FY2013-2022 speciation subset (n=14, mean iAs 20.6 ppb), TDS data (n=47), achievability percentages from Tables 3-4, modeled disease rates, the 42% exposure-reduction figure, and the historical 2008 23 ppb level of concern.
• Strengthened Methods to name ICP-MS (total) and HPLC-ICP-MS (speciation), LOQs, consumption inputs (NHANES, WWEIA), and the Taiwanese dose-response basis.
• Tightened the opening prose and Limitations to reflect the cancer + neurodevelopmental dual basis and the conservative total-As substitution in disease-rate modeling.

Audit subagent (2026-05-19, fresh-context general-purpose) flagged two ⚠️ concerns; both verified against the source and applied:

• Check 3 ⚠️: Methods section attributed the analytical methods to “FDA EAM 4.7 series” — the PDF footnote 9 only links to FDA lab method PDFs (UCM377005.pdf for total As; media/86499/download for HPLC-ICPMS speciation) and does not literally name “EAM 4.7.” Corrected to “FDA in-house method (link cited in guidance footnote 9)” for both methods.
• Check 5 ⚠️: Implications → Certification bullet proposed a “market-ratcheting target near 5 ppb iAs” derived from achievability data, which crosses the Part 2 boundary between describing occurrence data and proposing HMTc threshold values. Rewrote the bullet to surface the FY2011 and FY2013-2022 achievability percentages as occurrence inputs to the Standards Workbench, and to state explicitly that threshold selection is a workbench decision per Part 2 / Part 19.

Checks 1, 2, 4 returned clean. No false-positive rejections this pass.

Page history

The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.