Insecticides in American Streams: Detection Rates for Five Pest Control Active Ingredients
Key result: Across 119,264 water samples from US streams between 2015 and 2025, imidacloprid was detected in 41.6 percent and at 59.8 percent of monitored sites. Fipronil was found in 17.0 percent, chlorpyrifos 2.7, bifenthrin 2.6 and permethrin 1.3. Detection rates show no rise over the decade once reporting conventions are handled.
Summary of results
- This report analyses 119,264 stream water samples from the Water Quality Portal, the joint USGS, EPA and National Water Quality Monitoring Council data service, covering 2015 to 2025, for five insecticides used in structural pest control.
- Imidacloprid is by far the most frequently detected, in 14,433 of 34,686 samples, or 41.6 percent, and at 2,265 of 3,786 stream sites.
- Fipronil is second, in 17.0 percent of samples and at 31.9 percent of sites.
- The three others are rarely found in stream water: chlorpyrifos in 2.7 percent, bifenthrin 2.6 and permethrin 1.3.
- Measured concentrations are low in absolute terms. Median measured concentrations were 20.0 nanograms per litre for imidacloprid and 3.07 for fipronil.
- There is no evidence of a rise. Fipronil’s detection rate stayed between 14 and 20 percent every year from 2015 to 2024, even as laboratories began measuring smaller amounts.
- This version corrects an error. Our first version read about 16,600 results that programmes had coded as not detected as if they were detections, which inflated several rates severalfold. The correction is explained in our audit of the national monitoring record.
- Detection cannot be attributed to structural pest control. All five compounds have agricultural, turf, veterinary or structural uses.
Why this analysis did not exist
In short: The monitoring data is public, free and large. No pest control industry publication appears to have analysed it.
The Water Quality Portal aggregates water quality results from hundreds of federal, state, tribal and local monitoring programmes. It is queryable without a key and returns results by compound, site type and date. Trade discussion of pesticide runoff generally proceeds without reference to it. Our companion analysis of what chemicals exterminators actually use covers the product side. This report covers what is measured in streams.
Detection frequency by compound
| Compound | Water samples analysed | Detections | Detection rate | Stream sites | Sites with a detection | Site detection rate |
|---|---|---|---|---|---|---|
| Imidacloprid | 34,686 | 14,433 | 41.6% | 3,786 | 2,265 | 59.8% |
| Fipronil | 23,763 | 4,044 | 17.0% | 2,491 | 794 | 31.9% |
| Chlorpyrifos | 34,050 | 923 | 2.7% | 3,007 | 377 | 12.5% |
| Bifenthrin | 21,497 | 567 | 2.6% | 1,655 | 216 | 13.1% |
| Permethrin | 5,268 | 68 | 1.3% | 898 | 55 | 6.1% |
| Total | 119,264 | 20,035 |
Source: Water Quality Portal (waterqualitydata.us), siteType=Stream, sample dates from 1 January 2015, retrieved 25 September 2026. Results coded as not detected in the qualifier or comment fields are counted as non-detects; programmes that never report a non-detect are excluded for that compound. Author calculation.
What it means. Imidacloprid stands apart. It is found in two of every five samples and at three of every five monitored sites. Fipronil is the only other compound found routinely. Bifenthrin and permethrin, two of the most widely used structural perimeter insecticides, are found in fewer than three samples in a hundred. The most likely explanation is chemistry rather than volume: imidacloprid is water soluble and mobile, while pyrethroids such as bifenthrin and permethrin bind strongly to soil and sediment, which most water sampling does not measure.
What limits it. Monitoring is not randomly distributed across US streams: agencies sample where they expect to find something. These figures describe the sampled streams, not all streams. Site detection rates exceed sample rates because a site counts as positive if any single sample was positive.
Measured concentrations
| Compound | Measured results | Median (ng/L) | 90th percentile (ng/L) | Maximum (ng/L) |
|---|---|---|---|---|
| Imidacloprid | 14,423 | 20.0 | 98.3 | 25,000 |
| Fipronil | 3,994 | 3.07 | 11.6 | 382 |
| Chlorpyrifos | 885 | 5.4 | 94.4 | 4,200 |
| Permethrin | 68 | 4.6 | 28.0 | 55 |
| Bifenthrin | 567 | 3.0 | 52.4 | 840 |
Source: as above. Values converted to nanograms per litre. Detection limits entered as values are excluded.
What it means. One nanogram per litre is one part per trillion. The spread matters more than the median: the maximum imidacloprid result is more than 1,000 times the median, which indicates that a few samples captured episodic events such as storm runoff. Whether these values matter ecologically depends on compound-specific benchmarks, which we compare in a separate report.
What limits it. Concentration alone is not risk. Benchmarks differ by compound by several orders of magnitude.
What happened to the apparent surge
In short: Our first version reported that fipronil detections doubled in 2024 and 2025 and attributed it to more sensitive laboratories. On the corrected data there is no surge. The jump came from results that were coded as not detected.
| Year | Fipronil detection rate | Fipronil median measured (ng/L) | Imidacloprid detection rate | Bifenthrin detection rate |
|---|---|---|---|---|
| 2015 | 17.8% | 2.40 | 37.7% | 1.4% |
| 2016 | 15.6% | 4.20 | 35.0% | 2.8% |
| 2017 | 14.3% | 4.55 | 38.3% | 1.9% |
| 2018 | 18.6% | 3.55 | 51.2% | 4.4% |
| 2019 | 17.7% | 4.00 | 51.6% | 3.1% |
| 2020 | 17.1% | 2.83 | 48.4% | 5.9% |
| 2021 | 16.6% | 3.14 | 40.7% | 5.7% |
| 2022 | 18.3% | 2.69 | 40.2% | 1.6% |
| 2023 | 19.8% | 2.06 | 36.1% | 0.7% |
| 2024 | 17.4% | 0.79 | 35.3% | 1.4% |
| 2025 | 10.8% | 0.71 | 35.6% | 0.9% |
Source: as above, by calendar year. 2025 is a partial year. Full series for all five compounds, with the dominant laboratory method each year, in the CSV.
What it means. Fipronil’s detection rate sat between 14 and 20 percent every year from 2015 to 2024. Its median measured concentration fell by almost two thirds between 2023 and 2024, as the dominant laboratory method in the data changed, so laboratories were measuring smaller amounts; but that did not raise the detection rate. Imidacloprid peaked at about 51 percent in 2018 and 2019, was 48 percent in 2020 and has been between 35 and 41 percent since 2021. None of the five shows a sustained rise.
What went wrong in our first version. Indiana’s results, which enter the record in 2024 for fipronil and imidacloprid, code every result as “Present Below Quantification Limit” and never as not detected; Florida’s programmes enter detection limits as values and mark them as not detected only in the comment text. Read as detections, they produced a 2024 surge to 57 percent for fipronil and 34 percent for bifenthrin that does not exist on the corrected data.
What limits it. The network of contributing programmes changes from year to year, so even corrected annual rates are not a controlled trend. A defensible trend would hold the network and method constant.
Derived measures
1. Site Detection Ratio
Definition. The share of monitored stream sites at which a compound was detected at least once.
Formula. Distinct monitoring locations with at least one detection divided by distinct monitoring locations with at least one analysed sample.
Result. Imidacloprid 59.8 percent, fipronil 31.9 percent, bifenthrin 13.1 percent, chlorpyrifos 12.5 percent, permethrin 6.1 percent.
Limitations. A site with one sample and a site with 400 samples are treated identically.
2. Sensitivity Drift Index
Definition. The ratio of a compound’s median measured concentration in an early year to its median in a later year, used to flag changes in laboratory sensitivity.
Result. Fipronil 2023 to 2025: 2.06 / 0.71 = 2.9 times. Bifenthrin 2019 to 2023: 4.2 / 1.1 = 3.8 times. Imidacloprid 2016 to 2025: 26.5 / 9.4 = 2.8 times.
Interpretation. Laboratories now report smaller concentrations than they did, but on the corrected data this has not translated into higher detection rates. Any analysis that reports rising detections from this record should first check how non-detects were coded.
Limitations. Median concentration can also fall for environmental reasons, such as reduced use.
3. Mobility Inversion
Definition. The gap between a compound’s prominence in structural pest control use and its stream detection rank.
Observation. In California structural pest control in 2022, bifenthrin was used in 3,264,236 applications against fipronil’s 1,122,487, according to our analysis of structural pesticide use. In stream water, fipronil is detected in 17.0 percent of samples and bifenthrin in 2.6 percent. The more heavily applied compound is detected six times less often.
Limitations. The usage figures are California only and the detection figures national, so this is a suggestive comparison. Pyrethroids partition to sediment, which this water-sample analysis does not cover.
Related: fipronil’s breakdown products add detections a parent-only test misses; see testing for fipronil alone misses one in four detections.
Related: how these concentrations compare with EPA’s aquatic life benchmarks is in insecticides in US streams against EPA’s benchmarks.
Related: who uses these insecticides in California, and where they are found, in fipronil use and California streams.
Related research: insecticides in US groundwater.
Download the data
The tables in this report are published as CSV files under a Creative Commons Attribution 4.0 licence. You may republish, remix and build on them, including commercially, with attribution to The Pest Control Brief and a link to this page.
- Detection rates, site rates and concentrations for five compounds (CSV, 5 rows)
- Detection rate, median concentration and dominant laboratory method by compound and year (CSV, 55 rows)
How to cite this report
The Pest Control Brief, “Insecticides in American Streams: Detection Rates for Five Pest Control Active Ingredients”, 2 October 2026. https://thepestcontrolbrief.com/insecticides-in-us-streams/
References
- Water Quality Portal, a cooperative service of the US Geological Survey, the US Environmental Protection Agency and the National Water Quality Monitoring Council. Result search, narrowResult profile, siteType=Stream, from 1 January 2015, for imidacloprid, fipronil, chlorpyrifos, bifenthrin and permethrin. Example: waterqualitydata.us Result search, Fipronil. Retrieved 25 September 2026.
- Florida Department of Environmental Protection. Rule 62-160.700, Table 1: Data Qualifier Codes. floridadep.gov (PDF). Accessed 3 October 2026.
- US Environmental Protection Agency. WQX domain values: ResultMeasureQualifier. cdx.epa.gov ResultMeasureQualifier.CSV. Accessed 3 October 2026.
- California Department of Pesticide Regulation. Pesticide Use Report 2022. files.cdpr.ca.gov/pub/outgoing/pur_archives/. Source of the application counts.
Materials and methods
Source. 166,872 raw stream results for the five compounds, of which 47,608 were excluded as sediment, tissue, quality-control recoveries, passive-sampler units, unparseable values, results before 2015, or results from programmes that never report a non-detect for that compound. 119,264 water results were analysed.
Detection rule. Non-detects are results whose detection condition says so, and numeric results whose qualifier code contains the WQX code U or whose comment text carries Florida’s U flag. Detections are results marked “Present Below Quantification Limit” or “Trace”, and other positive numeric results. Programmes with at least 50 results for a compound, no non-detects and at least half below-quantification results are excluded for that compound: Indiana for all five, South Dakota for imidacloprid and fipronil, a volunteer watershed association for imidacloprid, and Missouri and Mississippi for chlorpyrifos.
Units. Micrograms per litre multiplied by 1,000; parts per trillion treated as nanograms per litre.
Limitations. Sampling is purposive. Reporting limits vary by laboratory. The network changes over time.
Corrections. First published 25 September 2026. On 3 October 2026 the report was rewritten after we found that about 16,600 results coded as not detected outside the detection-condition field had been read as detections. Every detection rate fell: imidacloprid from 53.3 to 41.6 percent, fipronil from 27.4 to 17.0, chlorpyrifos from 17.0 to 2.7, bifenthrin from 11.0 to 2.6 and permethrin from 39.4 to 1.3. The earlier finding of a 2024 to 2025 detection surge, and its attribution to laboratory sensitivity, was withdrawn; on the corrected data there is no surge. An earlier note about permethrin’s sensitivity to one programme was also withdrawn, because that sensitivity was produced by the same coding problem. See our corrections policy.
Last reviewed: 3 October 2026.
For reporters covering pesticides in water
Most quotable statistics
- Imidacloprid was detected in 41.6 percent of 34,686 US stream water samples from 2015 to 2025, and at 59.8 percent of monitored sites.
- Fipronil was detected in 17.0 percent of US stream samples.
- Bifenthrin and permethrin, widely used perimeter insecticides, were detected in 2.6 and 1.3 percent of samples.
- Fipronil’s detection rate stayed between 14 and 20 percent every year from 2015 to 2024.
- Bifenthrin was used in three times as many California structural applications as fipronil but is detected six times less often in US streams.
Caveats to carry with them
Detection is not harm. Monitored streams are not a random sample. These compounds have uses beyond structural pest control.
Press summary
Imidacloprid is found in two of every five US stream water samples, far more often than other insecticides used in pest control, according to an analysis of 119,264 samples in the national water-quality record by The Pest Control Brief. Fipronil is found in 17 percent; bifenthrin, chlorpyrifos and permethrin in fewer than 3 percent. Detection rates show no sustained rise between 2015 and 2025. The analysis corrects an earlier version that read results coded as not detected as detections.
Headlines editors could use
- Imidacloprid is in two of every five US stream samples
- The perimeter sprays that rarely show up in stream water
- A decade of stream data shows no rise in pest control insecticides
- Why the most-used insecticide is the least-detected
Questions about this report
Which pest control insecticide is most common in US streams?
Imidacloprid, detected in 41.6 percent of US stream water samples from 2015 to 2025, followed by fipronil at 17.0 percent.
Is bifenthrin found in streams?
Rarely in stream water: 2.6 percent of samples. Bifenthrin binds to sediment, which most water sampling does not measure.
Are insecticide levels in streams rising?
Not on these data. Corrected detection rates for all five compounds show no sustained rise from 2015 to 2025.
Why did an earlier version report higher numbers?
It read about 16,600 results that programmes had coded as not detected, in fields other than the detection condition, as detections. The correction is explained in our monitoring audit.
How much imidacloprid is in streams?
The median measured concentration was 20.0 nanograms per litre, or parts per trillion; the 90th percentile 98.3.
Does this mean pest control is polluting streams?
These data cannot attribute detections to any use. All five compounds have agricultural, turf or veterinary uses as well as structural ones.
Can I reproduce this?
Yes. Download the results from the Water Quality Portal and apply the detection rule in the methods.
