Methodology & Sources

Reported by Brian Martucci, Minnesota Reformer, 19 October 2026: “How much PFAS is in Minnesota fish? We often have no idea.”

Everything here comes from public state and federal records. This page explains exactly how the map is built, what it claims, what it does not claim, and how to check it. Figures computed 2026-10-30; last updated 2026-11-02.

Monitoring-plan facilities

MPCA also publishes PFAS groundwater results from its PFAS Monitoring Plan for permitted facilities such as operating landfills, airports and industrial sites (spreadsheet p-gen1-22i). These facilities are not on the Contamination Atlas, and MPCA marks most of their results “Prelim” (still being verified). Waters near them are screened with the same rules and exclusions as the cleanup sites, including MPCA’s statewide fish file. They are drawn on the map in orange and counted in the headline on their own line, so the cleanup-site figures can always be read alone. These figures will change if MPCA revises its preliminary results.

Surface water: what has been checked so far

Beyond fish, this project checked whether the water itself has been tested for PFAS near these sites. The figures below cover 2,057 lakes and river stretches: the 2,002 in the finding plus 55 that left it because their fish have a PFAS result in MPCA’s statewide file. Waters with a PFAS fish result published before this check (in MDH’s guidance, MPCA and MDH’s list since 2017, or another published study) are not among them.

From the data checked so far, 74 of the 2,057 have any PFAS result, fish or water (3.6%). That number could change once MPCA hands over the full export of PFAS results in its Surface Water Data, which it is preparing at this project’s request. When it arrives, these figures will be updated and the changes posted here.

• About 1,350 have been tested for something — fish for mercury or PCBs, or lab water samples. About 5.5% of those have a PFAS result.
• 429 have had their fish tested for mercury or PCBs (222 lakes sampled by MDH and 207 river stretches MPCA lists as impaired for mercury or PCBs in fish, a floor). 362 of them (84%) have no PFAS result of any kind, fish or water.
• 392 lakes have MPCA lab water samples since 2017 (nutrients, chlorophyll and similar). 390 of them have no PFAS water result.
• 9 have a PFAS water result from 2017 or later. Four of those are 2017 National Park Service samples from lakes in Voyageurs National Park, with no PFAS detected.

Where this project has checked

• MPCA’s Surface Water Data: its PFAS project downloads, plus station-by-station downloads for most screened lakes that have an MPCA station and a small share of river stretches
• The national Water Quality Portal: every Minnesota PFAS result from every agency, including EPA, the National Park Service and USGS
• MPCA’s statewide PFAS fish-tissue file (received 28 October 2026)
• MPCA and MDH’s list of PFAS detected in fish since 2017
• MDH’s fish consumption guidance for lakes and rivers
• MPCA’s 2024 and draft 2026 impaired waters lists
• EPA’s 2022 National Lakes Assessment and its National Rivers and Streams Assessment
• Published studies: EPA’s 2010 study (Delinsky et al.), MPCA’s 2007 metro lakes data and 2010 Pool 2 study, and the City of Sunfish Lake’s 2024 study

MPCA’s public surface-water site returns a captcha to automated downloads, so river stretches in particular have not all been checked station by station. Some PFAS surface-water results are known to sit outside MPCA’s public database — for example, PFAS that MPCA reported in a creek next to Gofer Landfill in 2021.

What this does not claim

Sources

Data as retrieved on the dates shown on each record. Agencies revise these datasets; the figures here are a snapshot as of the date at the top of this page.

How the finding is computed

1. Take every MPCA monitoring well with a recorded exceedance (exceedance_count > 0) whose exceedance names a PFAS compound. 8,543 wells on MPCA’s Contamination Atlas qualify statewide, plus 206 wells at facilities in MPCA’s PFAS Monitoring Plan (spreadsheet p-gen1-22i), most of their results marked preliminary.
2. Place each well in its USGS HUC12 sub-watershed, then follow the tohuc field up to two steps downstream.
3. Match every lake by its full DNR outline and every MPCA-listed stream stretch by its full line. A lake counts if more than 1 acre of it lies in a screened sub-watershed (a site’s own, or one or two downstream); a stretch counts if more than 50 feet does. Smaller overlaps are within mapping error (changed 30 October 2026).
4. Keep a waterbody only if MDH has never named a PFAS compound for it and MPCA has not listed it PFOS-impaired — i.e. there is no public PFAS result for its fish.
5. Record, for each one, which well, what it exceeded, by how much, and when it was last sampled.

How "tested for PFAS" is decided — and why it favors the agencies

A lake is treated as already tested for PFAS if MDH's guidance page names any PFAS compound at all, or if MPCA has listed it PFOS-impaired, or if it appears on MPCA and MDH’s published list of PFAS detections in fish since 2017, or if another published study or report gives a PFAS fish result for it (added 7 October 2026), or if it has a PFAS fish result in MPCA’s statewide PFAS fish-tissue file (added 30 October 2026). Lakes whose only published PFAS fish results are from 2007–2010 are listed separately. In practice that usually means a single compound — PFOS — while the PFAS family contains thousands. A lake screened only for PFOS is counted here as tested and is excluded from the finding, even though the compounds MPCA is measuring in groundwater at these sites routinely include PFOA, PFHxS, PFHxA, PFBS and PFNA as well.

That choice makes the count smaller and the claim narrower. It is deliberate: for every lake in this finding, no PFAS result for its fish has been published. Whether any were tested and found low is a question only MDH can answer.

Correction, 26 September 2026

A check against MDH’s own published lists found that this project’s page reader had missed PFOS guidance on 13 lakes. Seven were counted in the finding: Peltier, Centerville and Spring (Anoka County); Agnes, Henry and Winona (Douglas County); and Zumbro (Olmsted County). The advisory text had been captured correctly but was not recorded as PFAS. Those lakes are now excluded. Every lake’s saved guidance text is now checked directly for PFAS, in addition to the parsed table, and cross-checked against MDH’s April 2025 PFAS update table and March 2026 river guidelines. Those seven are removed. No river figures changed.

Correction, 27 September 2026

Lakes were previously de-duplicated by name within a county. That merged separate lakes that share a name (Minnesota has many “Mud,” “Long” and “Unnamed” lakes) and failed to merge basins of one lake that carry different names (Lake Vermilion was counted as four lakes). Lakes are now identified by DNR lake number: one lake per number, the whole-lake record preferred, and a lake is excluded if any of its basins has PFAS guidance. Against the 3 September figures: 42 basin records merged into their lakes, 7 lakes removed for PFAS guidance, and 145 separate lakes restored. The finding is now 1,031 lakes and 740,366 acres. River figures are unchanged.

Correction, 3 October 2026

The “Largest lakes” table below, and the measured values in map pop-ups, could show monitoring-well results for substances other than PFAS (for example manganese or boron) next to a PFAS label. They now show PFAS compounds only, each labelled with its compound. The map’s summary text was also brought in line with this page. For the 822 lakes whose MDH page says “Fish from this lake have not been tested for contaminants,” the map now quotes MDH directly. For the 209 lakes MDH has sampled, it now says “no published PFAS fish result” rather than “never tested,” because MDH’s guidance for those lakes names mercury or PCBs and does not say whether PFAS was checked. The map also separates lakes downstream of a site from lakes that share its sub-watershed, and no longer shows MPCA’s raw exceedance field as a percentage. No counts changed.

Correction, 7 October 2026

A reader pointed out that MPCA and MDH publish a list of PFAS detections in fish by waterbody since 2017 (“PFAS chemicals detected in fish by waterbody since 2017”). This project had not checked the finding against it. 18 lakes in the finding are on that list, so a PFAS result for their fish has been published, and this page was wrong to say otherwise for them. They have been removed from the finding and are listed below. All 18 were among the lakes MDH had sampled; none were among the lakes whose MDH page says “Fish from this lake have not been tested for contaminants.”

The finding changed from 1,031 lakes (740,366 acres; 209 sampled by MDH; 492 downstream of a site) to 1,013 lakes (706,924 acres; 191 sampled; 484 downstream). The number of lakes whose MDH page says they have not been tested is unchanged (822). The number of sites (94) and counties (60) did not change. Lakes on the list are now excluded automatically, by DNR lake number.

River reaches were checked the same day. The list identifies river samples by MPCA monitoring station, so each station was located in MPCA’s station records and matched to the reach it sits on. 9 reaches (92.0 miles) were removed: either a station on the list sits on the reach, or a sample with no reach of its own on file (a reservoir listed by DNR lake number, or a station without one) lies within 0.5 km of it. Stations on a different reach nearby, such as a tributary or the next reach, were not counted. River figures changed from 509 reaches and 4,305 miles to 500 reaches and 4,213 miles; facilities with lakes or rivers in the finding went from 110 to 109. 141 of the list’s samples are from Mississippi River pools; this finding included no Mississippi River reaches at the time (since 30 October 2026 it includes some; see that correction).

Correction, 7 October 2026 (second)

After the correction above, a wider search for other published PFAS fish results found 9 more lakes in the finding:

• Four of the state’s largest walleye lakes: Lake of the Woods, Mille Lacs, Vermilion and Cass. MDH and DNR tested them for PFCs (the older name for PFAS) in 2010. The only public record found is a May 2011 MPR report quoting an MDH scientist: “We didn’t find PFCs.” No per-lake results have been found; they have been requested from MDH.
• Independence, Sarah and Colby, in MPCA’s published 2007 metro lakes data, and Lake Byllesby, in a 2010 EPA study of fish collected in 2007.
• Sunfish Lake (Dakota County), tested by the City of Sunfish Lake in 2024, with all results below 10 ng/g. MDH’s page for this lake says its fish have not been tested.

Sunfish Lake has been removed. The 8 lakes tested only between 2007 and 2010 are now listed separately instead of being counted as having no published result. Those tests used older methods — MPCA’s 2007 data had a detection limit of about 5 ng/g, and about 10 compounds were measured, compared with about 40 today — and they predate MDH’s current health values. No later PFAS fish result has been published for them.

The finding changed from 1,013 lakes (706,924 acres) to 1,004 lakes (214,992 acres): 183 sampled by MDH (was 191), 821 whose MDH page says they have not been tested (was 822), and 479 downstream of a site (was 484), near 93 sites (was 94) in 59 counties (was 60). Another 8 lakes (491,885 acres) are listed separately as tested only between 2007 and 2010. River figures did not change.

Since checked: EPA’s 2022 National Lakes Assessment fish results (none of its Minnesota lakes is in the finding) and the full lake list of the 2010 EPA study (see the correction dated 8 October 2026). MPCA also reported that fish sampled in 2006 from the upper St. Croix River, above the St. Croix Falls dam, had PFOS below detection (MPCA, Mississippi River Pool 2 Intensive Study, 2010). The report does not say where on the river the fish were taken, so the finding’s two St. Croix reaches, both above the dam (Pine and Chisago counties, 59.7 miles), remain.

Correction, 8 October 2026

The supporting information of the 2010 EPA study cited above (Delinsky et al.) lists all 59 Minnesota lakes whose fish it measured for PFAS, from fish collected in 2007. 7 of them were still in the finding: Gull, Koronis, Embarrass, Mina, Hammal, Hanging Kettle and Winona. All had been sampled by MDH, whose guidance for them names mercury or PCBs. None is among the lakes whose MDH page says they have not been tested. Under the exclusion rules, they are now listed separately with the other lakes tested only between 2007 and 2010.

The finding changed from 1,004 lakes (214,992 acres) to 997 lakes (200,010 acres): 176 sampled by MDH (was 183), 821 whose MDH page says they have not been tested (unchanged), and 477 downstream of a site (was 479), near 93 sites (unchanged) in 59 counties (unchanged). 15 lakes (506,868 acres) are now listed separately as tested only between 2007 and 2010. River figures did not change.

For context: of the 15 lakes now listed separately, 10 had no PFAS reported (the four walleye lakes, per MDH in 2011, and six of the seven above), and 5 had PFOS between 1.4 and 22 ng/g (Byllesby, Winona, Independence, Sarah and Colby). These results are from 2007 to 2010, before MDH’s current health values, and no later PFAS fish result has been published for these lakes.

Source: Delinsky et al. 2010, Environ. Sci. Technol. 44:2549 (EPA), Supporting Information Tables S1 and S4.

Correction, 30 October 2026

Five changes, all applied the same day.

1. MPCA’s statewide PFAS fish file. On 28 October MPCA sent its statewide PFAS fish-tissue results (163,898 results from 2004 to August 2025, including non-detects). 23 lakes and 20 river stretches in the finding have PFAS fish results in that file or in EPA’s national river surveys, and have left the finding. 18 of the lakes are removed; 5 (Loon, Oak, Penn, Louise and Green Mountain) were tested only between 2007 and 2010 and are listed separately with the other older-test lakes. Two of the 23 lakes, Wabasso and Placid, are lakes whose MDH page says their fish have not been tested.

2. River miles are no longer reported. This page, and the Minnesota Reformer’s 19 October story, gave about 4,200 river miles. That figure added the full length of every stretch on the list, but often only part of a stretch lies in a screened sub-watershed, and the total depended heavily on how partial stretches were counted. River stretches are counted instead.

3. Matching uses full shapes. Lakes are now matched by their full DNR outline and river stretches by their full MPCA line, instead of a center point and three points along each stretch. A lake counts if more than 1 acre of it lies in a screened sub-watershed; a stretch counts if more than 50 feet does. Smaller overlaps are within mapping error between the two map layers. The lake set did not change (997). Of the 500 river stretches, 497 were the same, 3 dropped out as slivers and 3 came in.

4. River advice was applied by name, not by stretch. MDH’s PFAS fish advice for one part of a river was removing every stretch with that river’s name: all of the Mississippi (MDH’s advice covers Pools 2 through 6), all of the Crow River (the advice covers only the South Fork in Carver and McLeod counties), and a Miller Creek in Wabasha County (the advice is for a different Miller Creek in St. Louis County). Advice is now applied only to the stretches it covers. 22 stretches were restored: 11 on the Mississippi above Minneapolis and below Pool 6, 10 on the Crow River and its North and Middle forks, and Miller Creek. A twelfth Mississippi stretch (Pool 9) stays out because MPCA’s 2023 list includes PFAS fish from it.

5. Monitoring-plan facilities added. Waters near permitted facilities in MPCA’s PFAS Monitoring Plan are now screened with the same rules and counted in the headline on their own line: 332 lakes and 194 river stretches near 52 facilities, where MPCA marks most results preliminary. The cleanup-site figures below are not affected.

6. Five lakes had newer PFAS fish results than the older-tests list showed. Mille Lacs (2024, PFOS not detected), Lake of the Woods (2023), Vermilion (East basin, 2024), Gull (2023) and Embarrass (2023) were listed as tested only between 2007 and 2010. MPCA’s statewide fish file has newer PFAS fish results for all five, all low (average PFOS 1.3 ng/g or less). They are no longer listed as older tests; 15 lakes remain on that list.

River stretches step by step: 500 before these changes; 500 after full shapes (3 out, 3 in); 522 after the river-advice fix (22 restored); 502 after MPCA’s fish file (20 removed).

The finding changed from 997 lakes (200,010 acres) and 500 river stretches to 974 lakes (189,298 acres) and 502 river stretches: 155 sampled by MDH (was 176) and 819 whose MDH page says they have not been tested (was 821), near 93 sites in 59 counties (unchanged). 20 lakes are now listed separately as tested only between 2007 and 2010 (was 15).

Source: MPCA statewide PFAS fish-tissue file (sheet AllFish.10282026), received from MPCA on 28 October 2026; EPA National Rivers and Streams Assessment via the Water Quality Portal for six stretches.

How the exceedance figures work

MPCA's well records carry a derived field, exceedance_pct_max. Despite the name it is a multiple of the health value, not a percentage, and it is computed against the most protective value MDH publishes for that compound.

Worked example — Hickory Grove Landfill, well 785553. MPCA reports a PFOA result of 0.143 µg/L and a field value of 18,101.27. MDH's Human Health-Based Water Guidance Table gives two values for PFOA:

So MPCA's figure is arithmetically correct and refers to the cancer-based value. A PFOS well at the Lehillier / Mankato site confirms the same pattern: 0.007 µg/L against MDH's chronic PFOS value of 0.0023 µg/L gives 3.04, which is exactly what MPCA reports.

Because a bare multiple does not say which value it is measured against, this site shows the measured concentration first and then both multiples, each labelled. Every figure can be checked against MDH's guidance table without trusting arithmetic in between.

An earlier version of this page said MPCA's field was inflated by a factor of 1,000. That was wrong — it came from working backwards to a standard rather than reading MDH's published table — and was corrected on 2026-09-03 before publication. The raw field is retained in the downloadable data as mpca_exceedance_pct_max_RAW_see_methodology.

Exclusions applied

• Program rollups. Some MPCA well records carry a program name rather than a facility name (for example "East Metro PFAS", 5,585 wells spanning three degrees of latitude). Their wells are real; their location attribution is not. Excluded.
• Scrape errors. A page this project failed to read is not a lake that was never tested. Excluded rather than counted.
• Out of state and non-lakes. Waterbodies outside Minnesota, harbors, and river reaches listed in the lake table.
• Duplicate basins. Large lakes appear as several DNR basins. Lakes are identified by DNR lake number: one lake per number, the whole-lake record preferred, and a lake is excluded if any of its basins has PFAS guidance.
• Published PFAS fish results. A lake on MPCA and MDH’s list of PFAS detections in fish since 2017 already has a published result and is excluded (matched by DNR lake number; added 7 October 2026). A river reach is excluded when a station on the list sits on it, or when a sample with no reach of its own on file lies within 0.5 km of it.
• Other published PFAS fish results. Lakes with a PFAS fish result in another published source (an agency report, a peer-reviewed study, or a local government study) are excluded; those whose only results are from 2007–2010 are listed separately (added 7 October 2026).
• MPCA’s statewide fish file. A lake or river stretch with a PFAS fish result in MPCA’s statewide PFAS fish-tissue file is excluded; lakes whose only results are from 2007–2010 are listed separately (added 30 October 2026). MDH’s river advice is applied only to the stretches it covers.
• Stale readings. Wells last sampled before 2020 are flagged and left out of headline figures. Fresher wells claim a sub-watershed first.

Largest lakes MDH sampled, with no published PFAS result

Scale of the underlying gap

Minnesota has 8,628 lakes over 25 acres in the DNR hydrography layer. 1,480 have a fish-tissue result of any kind. 7,148 have none. Separately, DNR has published 21,933 fish population surveys covering 3,321 lakes, and MPCA has 19,599 monitoring wells in its contamination atlas.

Known limitations

• Surface watershed connection is not evidence of groundwater transport. Groundwater plumes follow aquifer flow, which this project did not model.
• MDH's statewide guidance applies to untested lakes and is built to be protective in the absence of waterbody-specific data. This map does not claim otherwise; it shows where that default is doing the work.
• Fish population comparisons use DNR's own normal ranges by lake class. Gear types are never mixed; survey methods changed over time and trend comparisons are restricted to standard surveys.
• A monitoring well often exceeds standards for several substances at once — manganese, boron, 1,4-dioxane and PFAS together. Only the PFAS compounds are shown on each record, and only a PFAS exceedance qualifies a waterbody for the finding.
• Some MPCA records are site-level rather than well-level and carry coarser locations.

Corrections

If something here is wrong, I want to fix it. Email [email protected] with the record and what is incorrect. Corrections are made within 48 hours and noted here. Agencies, facility operators, and anyone else named in this data are welcome to write.

This is an independent project. It is not affiliated with or endorsed by MDH, MPCA, DNR, EPA, USGS, or any company named in the underlying records.

← Back to the map