Health Canada 2006 — Guidelines for Canadian Drinking Water Quality: Arsenic

Health Canada’s May 2006 Guideline Technical Document for arsenic in drinking water, prepared with the Federal-Provincial-Territorial Committee on Drinking Water of the Federal-Provincial-Territorial Committee on Health and the Environment, establishes a maximum acceptable concentration (MAC) for arsenic in drinking water of 0.010 mg/L (10 µg/L) based on municipal- and residential-scale treatment achievability. The document classifies arsenic as a Group 1 human carcinogen and notes that the MAC is higher than the concentration that would represent an “essentially negligible” lifetime cancer risk (estimated at 0.3 µg/L on the basis of internal organ cancer endpoints in the southwestern Taiwan cohort). The document is identified as a risk-management decision: every effort should be made to reduce arsenic in drinking water to as low as reasonably achievable (ALARA), and the 10 µg/L MAC reflects the practical quantitation limit (3 µg/L), the reliable measurement limit (5 µg/L), and what municipal- and residential-scale treatment can practically achieve at reasonable cost, especially for smaller public systems and private wells. Lifetime cancer risk at the MAC is estimated at 3.0 × 10⁻⁵ to 3.9 × 10⁻⁴ — above the 10⁻⁶ to 10⁻⁵ range Health Canada defines as “essentially negligible.” The document compiles Canadian provincial and territorial drinking-water arsenic occurrence data from 1976 to 2002, reviews analytical methods and municipal/residential treatment technologies, and synthesizes the human and animal toxicology underlying the carcinogenicity classification.

This is the May 2006 guideline document, which supersedes prior Health Canada guideline technical documents on arsenic in drinking water (the 1989 guideline of 25 µg/L, lowered following review of the southwestern Taiwan internal-organ cancer evidence). The 10 µg/L MAC is numerically the same as the US EPA 10 µg/L MCL (see epa-arsenic-mcl) and the WHO 10 µg/L provisional guideline; the underlying derivations are not all identical and are described in the respective agency documents.

Key numbers

Guideline values (Sections 1.0 and 11.0):

Unit risk and lifetime cancer risk estimates (Section 10.0, Health Canada 2005 quantitative risk assessment using Poisson-Weibull model from Morales et al. 2000 fit to southwestern Taiwan data of Chen et al. 1985 and Wu et al. 1989):

Estimated lifetime range of risk of excess internal organ cancers at various drinking-water arsenic concentrations (Table 1, p. 3, 70-year lifetime, additional to background, × 10⁻⁵):

Canadian drinking-water occurrence data (Section 5.1, provincial and territorial monitoring submissions):

International occurrence context (Section 5.1):

Estimated mean daily intake of total arsenic from food (Section 5.2):

Inorganic-arsenic share of food and water intake (Section 5.5):

EPA-approved analytical methods for arsenic in drinking water (Table 2, p. 9):

Treatment performance (Section 7.1):

Health-effect anchoring values (Sections 9.1, 9.3):

Southwestern Taiwan cohort (Section 9.1, Chen et al. 1985, 1986, 1992; Wu et al. 1989; Tseng 1977):

Antofagasta, Chile cohort (Section 9.1, Zaldívar 1980; Zaldívar and Ghai 1980):

US studies cited as not finding clear associations at lower exposures (Section 9.1):

Methods (brief)

This is a federal-provincial-territorial guideline technical document, not a primary research study. The document is the May 2006 update of the Health Canada arsenic guideline, replacing prior guideline technical documents (the 1989 25 µg/L guideline was lowered to 10 µg/L following review of southwestern Taiwan internal-organ cancer evidence, US EPA 2001 review, and US NRC 1999 and 2001 reviews). The document was prepared by the Water Quality and Health Bureau, Healthy Environments and Consumer Safety Branch, Health Canada, and adopted by the Federal-Provincial-Territorial Committee on Drinking Water.

Risk-assessment derivation: Health Canada (2005) applied the Poisson-Weibull model recommended by US EPA (2001a) and fit by Morales et al. (2000) to southwestern Taiwan internal-organ cancer data of Chen et al. (1985) and Wu et al. (1989), using an external unexposed comparison population per US NRC (2001) recommendation. The external comparison population was the southwestern Taiwanese rural population rather than the entire Taiwanese population, to reduce bias from urban–rural differences. Assumptions followed US EPA (2001a): choice of risk metric, 1% extra-risk benchmark response, conversion factor from southwestern Taiwanese to Canadian drinking-water consumption and body weight. The most sensitive endpoint for males and females was lung cancer (3.85 × 10⁻⁵ at 1 µg/L); the lower bound of the unit risk range was liver cancer (3.06 × 10⁻⁶). The unit risk was multiplied by drinking-water arsenic concentrations to produce lifetime cancer risk estimates at 0.3, 5, 10, and 25 µg/L (Table 1).

The MAC was set by considering treatment achievability and the PQL alongside the health-based “essentially negligible” target concentration. The 0.3 µg/L health-based target falls within the 10⁻⁶ to 10⁻⁵ “essentially negligible” range Health Canada applies to carcinogens. The 10 µg/L MAC is acknowledged as a risk-management decision that exceeds the health-based value because it reflects what municipal- and residential-scale treatment can practically achieve and what laboratories can reliably measure. Cancer risk at the MAC (3.0 × 10⁻⁵ to 3.9 × 10⁻⁴) is therefore higher than the “essentially negligible” range, and the document directs ALARA application below the MAC.

Occurrence data were drawn from provincial and territorial monitoring submissions (Prince Edward Island Department of Fisheries, Aquaculture and Environment 2003; Ministère de l’Environnement du Québec 2003; Ontario Ministry of Environment and Energy 2003; Saskatchewan Department of Environment and Resource Management 2003; Alberta Department of Environment 2003; Nova Scotia Department of the Environment 1998; Newfoundland Department of Environment and Labour 2003) and from published Canadian groundwater studies (Méranger 1984; Boyle 1998).

Analytical-methods review summarized the EPA-approved arsenic-in-water methods and method detection limits. The PQL of 3 µg/L is the value EPA assigned for laboratories’ practical measurement capability under EPA-approved methods (US EPA 1999). EPA Method 1632 was singled out as providing direct GHAA speciation of arsenic, with an MDL of 0.002 µg/L, but is described as requiring a high degree of skill.

Treatment-technology review summarized municipal-scale (coagulation/filtration, lime softening, activated alumina, ion exchange, reverse osmosis, manganese greensand filtration, adsorption/filtration with iron-oxide-coated sand or granular ferric hydroxide, electrodialysis reversal) and residential-scale (reverse osmosis, distillation, adsorption/filtration) options, plus water blending as a non-treatment management approach. The document references NSF/ANSI Standards 53 (Drinking Water Treatment Units — Health Effects), 58 (Reverse Osmosis Drinking Water Treatment Systems) and 62 (Drinking Water Distillation Systems) as the relevant certification standards for residential devices; NSF/ANSI standards test for removal to 10 µg/L, and certified devices are often designed to remove arsenic to well below that concentration.

Toxicology synthesis drew on US EPA 2001a, US NRC 1999 and 2001, ATSDR 2000, WHO 2003, IPCS 2001, Health Canada 1998, IARC and Health Canada (1994) carcinogenicity reviews, the southwestern Taiwan cohort (Tseng 1977, Chen 1985, 1986, 1992; Wu 1989), the Chilean (Antofagasta) cohort (Zaldívar 1980; Borgono and Greiber 1971), Bangladeshi (Smith 2000), Indian (Mandal 1998), Mexican (Cebrian 1983), and US (Steinmaus 2003; Lamm 2004; US EPA and AwwaRF 2004; Aschengrau 1989; Zierler 1988) studies, plus animal studies in rats, mice, hamsters, rabbits and chicks (Carmignani 1985; Yamamoto 1995; Wanibuchi 1996, 1997; Hanlon and Ferm 1985, 1986a, 1986b; Hood and Bishop 1972; Zierler 1988; Ferm and Hanlon 1985; Nemec 1998). Genotoxicity, mode-of-action, and relative-toxicity-of-species data (Aposhian 2000; Petrick 2000; Styblo 2000; Mass 2001) were reviewed. Although the document notes recent evidence that MMAᴵᴵᴵ may be more toxic than inorganic species, the Panel concluded the mode-of-action evidence was not strong enough to justify a non-linear quantitative risk assessment model.

Implications

• Reference value: Health Canada’s 2006 MAC of 10 µg/L (0.010 mg/L) for total arsenic in drinking water is a risk-management decision based on treatment achievability and the practical quantitation limit, not on the health-based “essentially negligible” target of 0.3 µg/L. Lifetime cancer risk at the MAC is estimated at 3.0 × 10⁻⁵ to 3.9 × 10⁻⁴, above the 10⁻⁶ to 10⁻⁵ “essentially negligible” range. The Health Canada derivation uses the Morales et al. 2000 Poisson-Weibull dose-response fit to the southwestern Taiwan internal-organ cancer data of Chen et al. 1985 and Wu et al. 1989, per US EPA 2001a’s recommendation. The 10 µg/L MAC is numerically the same as the US EPA 10 µg/L MCL (see epa-arsenic-mcl) and the WHO 10 µg/L provisional guideline, but the underlying derivations of those values are not described in this document and should not be presumed identical.
• Speciation discipline: the MAC is set on total arsenic, not inorganic arsenic, despite the document explicitly framing carcinogenic risk on inorganic forms and noting that drinking water arsenic is essentially all inorganic per the US NRC (1999) review. The total-arsenic basis simplifies compliance monitoring; speciation is described as available (EPA Method 1632, HPLC-GHAA, HPLC-ICP-MS) but not required for compliance.
• ALARA framing: the document explicitly directs that arsenic levels in drinking water should be maintained as low as reasonably achievable below the MAC. Residential-scale certified treatment devices reduce arsenic to well below 10 µg/L, and combined-system designs can achieve below 5 µg/L.
• Occurrence context: provincial and territorial monitoring across 1976–2002 indicates Canadian drinking-water arsenic is generally well below the MAC (>97–99% of samples below 10 µg/L in most provinces’ aggregate data), with elevated levels concentrated in well water in areas of natural geological sources (Nova Scotia, Newfoundland, Saskatchewan, Alberta, British Columbia, parts of Ontario). Maximum observed levels reach 105 µg/L (Saskatchewan), 580 µg/L (Bowen Island, BC), 368 µg/L (Newfoundland school wells), 288 µg/L (Newfoundland public wells), and 1000 µg/L (Alberta).
• Dietary context: drinking water is not the major source of arsenic exposure for most Canadians; food is the primary route, principally seafood (78.9% of dietary arsenic intake in a US survey; 94% of total arsenic intake in the 1997 UK MAFF total diet study). However, the inorganic-arsenic share is higher in drinking water than in seafood (where the dominant species is arsenobetaine), so the inorganic-arsenic contribution from drinking water can rival that from food.
• Subgroup framing: Health Canada explicitly states that subpopulations (children, pregnant women) are not at greater risk of developing health effects from arsenic exposure than the general population in the 2006 derivation (Sections 2.1, 3.0). The document references the US NRC conclusion that “there was insufficient scientific information to permit separate cancer risk estimates for potential subpopulations such as pregnant women, lactating women and children” (US EPA 2001a, cited Section 9.1).

Wiki pages this source may touch

• arsenic
• arsenic-inorganic
• arsenic-total
• water
• epa-arsenic-mcl — international comparator (US EPA 10 µg/L MCL on which the Canadian MAC was harmonized)
• efsa-arsenic-contam-2009 — international comparator (EFSA 2009 BMDL01 range 0.3–8 µg/kg bw/day)
• jecfa-arsenic-bmdl — international comparator (JECFA arsenic reference points)

Verification notes

• Source type is regulatory-guideline: this is a Health Canada Federal-Provincial-Territorial guideline technical document (May 2006). The document supersedes prior Health Canada guideline technical documents on arsenic in drinking water. No DOI was assigned (the document predates routine DataCite minting for Health Canada technical documents). Suggested citation per p. ii: “Health Canada (2006) Guidelines for Canadian Drinking Water Quality: Guideline Technical Document — Arsenic. Water Quality and Health Bureau, Healthy Environments and Consumer Safety Branch, Health Canada, Ottawa, Ontario.” License: Government of Canada reproducible for personal/internal use with attribution.
• metals: [iAs, tAs]. The MAC is set on total arsenic (so tAs is the analyte the regulation polices), but the document explicitly frames the carcinogenic risk on inorganic arsenic and notes that drinking-water arsenic is “essentially all inorganic” per US NRC 1999; the underlying dose-response model is therefore an inorganic-arsenic model. Both abbreviations apply per Part 14 speciation discipline.
• matrices: [drinking-water]. The document covers Canadian municipal and non-municipal drinking water (surface water, groundwater, treated and raw, public utilities and private wells). The drinking-water umbrella matrix slug is used per Part 5b routing-layer fan-out guidance.
• products: [] is correct: the document describes drinking water as a regulatory matrix, not a packaged product category. The bottled-drinking-water and mineral-water product slugs exist but are not the subject of this guideline.
• ingredients: ["[[ingredients/water]]"] covers the water-as-ingredient routing. Other foods (fish, seafood, vegetables) are mentioned only in the dietary-exposure benchmarking context and are not the subject of the guideline.
• jurisdictions: [CA] covers all 9 sampling provinces/territories listed in sampling_locations (the Yellowknife air monitoring is incidental). Canadian national-level guideline.
• Two PDFs at raw/Manual Fetch Kimi /June 3 Folder/water-arsenic-eau-eng.pdf and raw/Manual Fetch Kimi /June 3 Folder/water-arsenic-eau-eng 2.pdf are bit-identical (same SHA-256 eb62f3177f0d1f01ff601ef670071f8220eee2023c649ba07019bcdbe213a5ce). The 2.pdf variant is being processed per the manual-fetch ingest tracker; the un-suffixed canonical name is recorded under near_duplicates.
• Regulatory-slug proposal (for Karen): a new regulations/health-canada-arsenic-mac.md slug would be justified for this document because it sets a binding Canadian MAC of 10 µg/L for total arsenic in drinking water and is the load-bearing Canadian counterpart to [[regulations/epa-arsenic-mcl]]. Per skill policy this skill does not create new regulation pages; surfaced here for the next regulation-slug review. The existing arsenic-regulation slugs include epa-arsenic-mcl, epa-iris-inorganic-arsenic-rfd, efsa-arsenic-contam-2009, eu-2015-1006-iAs-rice, eu2023-arsenic-rice-based-drinks, fda-iAs-rice-cereal-100ppb, jecfa-arsenic-bmdl — no Health Canada arsenic page exists yet.
• Part 12 brand firewall: source contains no brand-attributed contamination data. NSF/ANSI Standards 53, 58, and 62 are standards-body references, not commercial brand attributions; NSF International, Canadian Standards Association International, Water Quality Association, Underwriters Laboratories Inc., Quality Auditing Institute, and International Association of Plumbing & Mechanical Officials are listed as Standards Council of Canada-accredited certification organizations (regulatory-event/standards-body exception 1; not consumer-product brand attribution).
• Part 2 wiki/HMTc firewall: this page reports the Health Canada guideline values, derivation steps, occurrence data, and international comparators as the document presents them. The MAC’s explicit framing as a risk-management value above the health-based “essentially negligible” target is faithful to the source’s own framing — the document itself states the MAC “exceeds the health-based guideline value” and is a “risk management decision.” No HMT&C threshold comparisons, no consumer risk advisories, no proposals for HMT&C certification levels.
• Brand-firewall exception 1 (regulatory event subject) does not apply — the document does not name any branded products as the subject of a regulatory action. The “certified treatment devices” language is product-category, not brand-specific.
• Brand-firewall exception 2 (scientific method vendor names): no instrument vendor names are given in the document; EPA, APHA Standard Methods, and ASTM method numbers are cited as standards, not brands.
• 2026-06-04 — Audit subagent (general-purpose, fresh context) verdict REVISE. Two ⚠️ findings in Check 5 (Part 2 wiki/HMTc firewall), no ❌ findings. Checks 1 (numerical fidelity across MAC, unit-risk, Table 1, all provincial occurrence ranges, Table 2 MDLs, dietary intake, Taiwan/Antofagasta/Steinmaus/Lamm/EPA-AwwaRF figures, lethal-dose range, half-life, relative-toxicity order), 2 (slug vocabulary), 3 (speciation and methods), and 4 (Part 12 brand firewall) returned clean. Both ⚠️ findings independently verified against the PDF and applied:
  • “This contrasts with later assessments (US EPA, EFSA) that have applied subgroup-specific exposure scaling for children” removed from the Subgroup-framing Implications bullet — verified against PDF pp. 2–3, 24: source only states that subpopulations are not at greater risk, does not contrast its position with later EFSA/US EPA assessments. Replaced with the source-internal US NRC reference Health Canada itself cites (US EPA 2001a, §9.1).
  • “all built on the southwestern Taiwan internal-organ cancer evidence with the Morales et al. 2000 Poisson-Weibull dose-response fit” narrowed to the Health Canada derivation alone — verified against PDF pp. 22–23: source documents only the Canadian use of Morales 2000 (cited Section 10.0). WHO 2003 and US EPA derivations of the 10 µg/L limit are not described in this document; the “all built on” phrasing over-claimed methodological identity across three agencies. Top-paragraph sentence and Implications bullet both narrowed accordingly.

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.