What the AQI is and what it is not
The Air Quality Index is a translation device. Six criteria pollutants are measured in incompatible units — micrograms per cubic metre for particles, parts per million for ozone and carbon monoxide, parts per billion for the two gases — and the index maps each onto one 0-to-500 scale calibrated to health effects. That lets a single number carry a public health message, and it lets ozone at 0.078 ppm and PM2.5 at 35.9 µg/m³ be compared as roughly comparable hazards.
The anchoring is deliberate: for each pollutant, the concentration equal to the National Ambient Air Quality Standard maps to AQI 100. That is what makes the sub-indices commensurate. An AQI of 100 means the same regulatory thing regardless of which pollutant produced it, and it is why the category boundaries fall at 50, 100, 150, 200 and 300 for every pollutant rather than at round concentrations.
What the index is not: it is not a measure of total pollution, and it is not an average. It reports the worst sub-index, on the reasoning that the health message must reflect the most hazardous constituent present. Two pollutants each at AQI 90 report 90, not 180. It also says nothing about pollutants outside the six, including air toxics, pollen and radon.
The interpolation, step by step
Each pollutant has a breakpoint table of concentration bands paired with index bands. The calculation has three parts.
Truncate the concentration. Not round — truncate. PM2.5 goes to one decimal place, PM10 and the two gases in ppb to whole numbers, ozone to three decimals, carbon monoxide to one. A PM2.5 reading of 35.97 is treated as 35.9. This step is what makes the small gaps between bands — 9.0 and then 9.1, 35.4 and then 35.5 — unreachable, and it is why two monitors reporting the same air can differ by one AQI point.
Find the band and interpolate. Locate the band whose lower and upper concentration breakpoints bracket the truncated value, then interpolate linearly between the corresponding index values. The formula is exactly the equation of the straight line through the two corners of that band. Round the result to the nearest integer.
Take the maximum. Compute a sub-index for every pollutant you have data for, and report the largest. The pollutant producing it is the responsible or driving pollutant, and it is what the health message should refer to.
Because the mapping is piecewise linear rather than proportional, index differences are not concentration differences. Between AQI 0 and 50 each PM2.5 index point costs 0.18 µg/m³; between 151 and 200 each point costs about 1.4 µg/m³. The scale deliberately stretches at the clean end and compresses at the dirty end. A jump from AQI 150 to 200 is a much larger change in the air than a jump from 50 to 100.
One further practical note: the EPA uses a separate NowCast weighting for real-time particle reporting, which weights recent hours more heavily to approximate the 24-hour average before the day is over. That is why an app's current AQI can differ from what you get here by putting a single hourly reading in. Enter a genuine 24-hour average for the definitional answer.
Worked example: ozone at 0.078 ppm
A monitor reports a maximum daily 8-hour ozone average of 0.078 ppm. Nothing else is elevated.
- Truncate. Ozone is carried to three decimal places, so Cp = 0.078 ppm.
- Find the band. 0.078 lies in the 0.071–0.085 ppm band, which corresponds to AQI 101–150.
- Set the four corners. BPLo = 0.071, BPHi = 0.085, ILo = 101, IHi = 150.
- Slope. (150 − 101) ÷ (0.085 − 0.071) = 49 ÷ 0.014 = 3,500 index points per ppm.
- Interpolate. 3,500 × (0.078 − 0.071) = 3,500 × 0.007 = 24.5; add ILo: 24.5 + 101 = 125.5.
- Round. AQI 126, category Unhealthy for Sensitive Groups, driving pollutant ozone.
Now suppose PM2.5 is also elevated, at 35.9 µg/m³. Its band is 35.5–55.4 µg/m³ mapping to 101–150, so the slope is 49 ÷ 19.9 = 2.4623 points per µg/m³, and the sub-index is 2.4623 × (35.9 − 35.5) + 101 = 0.985 + 101 = 101.98, rounding to 102.
The overall AQI is still 126. Both pollutants are in the same health category, but the index reports the higher of the two rather than combining them, so the reported value and the driving pollutant are unchanged. That is the design of the index, and it is also its main limitation: simultaneous exposure to several pollutants near the same level is not treated as worse than exposure to one of them.
Reading the categories
Six categories divide the scale, and the boundaries are what matter rather than the exact number. 0–50 Good: air quality is satisfactory and pollution poses little or no risk. 51–100 Moderate: acceptable, with a risk for people unusually sensitive to the driving pollutant. 101–150 Unhealthy for Sensitive Groups: members of sensitive groups may experience health effects while the general public is less likely to be affected. 151–200 Unhealthy: some of the general public may experience effects, and sensitive groups more serious ones. 201–300 Very Unhealthy: a health alert, with increased risk for everyone. 301 and above Hazardous: a health warning of emergency conditions.
Who counts as a sensitive group depends on the driving pollutant, which is why that output matters as much as the number. For ozone and particles it includes people with asthma or other lung disease, people with heart disease, older adults, children and teenagers, and people active outdoors. For carbon monoxide it is people with cardiovascular disease. An AQI of 130 driven by ozone and one driven by PM2.5 call for advice aimed at overlapping but not identical populations.
Treat single-point differences as noise. Given the truncation step, monitor uncertainty and the choice of averaging window, an AQI of 98 and an AQI of 103 describe air that is not meaningfully different, even though they fall in different categories. Use the categories for decisions and the number for trends.
Finally, if you are working with a low-cost sensor, be aware that most report a raw particle estimate that has not been corrected against a reference instrument. Correction factors matter a great deal at the concentrations where the category boundaries sit, and an uncorrected sensor reading pushed through this calculator inherits whatever bias the sensor has.
AQI breakpoints by pollutant
| AQI band | Category | PM2.5 | PM10 | O₃ (8-h) | CO (8-h) | SO₂ (1-h) | NO₂ (1-h) |
|---|---|---|---|---|---|---|---|
| 0–50 | Good | 0.0–9.0 | 0–54 | 0.000–0.054 | 0.0–4.4 | 0–35 | 0–53 |
| 51–100 | Moderate | 9.1–35.4 | 55–154 | 0.055–0.070 | 4.5–9.4 | 36–75 | 54–100 |
| 101–150 | Unhealthy for Sensitive Groups | 35.5–55.4 | 155–254 | 0.071–0.085 | 9.5–12.4 | 76–185 | 101–360 |
| 151–200 | Unhealthy | 55.5–125.4 | 255–354 | 0.086–0.105 | 12.5–15.4 | 186–304 | 361–649 |
| 201–300 | Very Unhealthy | 125.5–225.4 | 355–424 | 0.106–0.200 | 15.5–30.4 | 305–604 | 650–1,249 |
| 301–500 | Hazardous | 225.5–325.4 | 425–604 | not defined | 30.5–50.4 | 605–1,004 | 1,250–2,049 |
PM2.5 breakpoints are those set by the EPA's 2024 revision of the fine particle standard, which lowered the Good/Moderate boundary from 12.0 to 9.0 µg/m³. If you are reproducing an older reported value, check which table it used.
Mistakes that produce the wrong AQI
- Averaging the sub-indices. The overall AQI is the maximum, not the mean. Averaging systematically understates the index whenever more than one pollutant is elevated.
- Rounding the concentration instead of truncating it. The method truncates. A PM2.5 value of 35.48 is 35.4, not 35.5, and that single step moves it across a category boundary.
- Using an hourly reading where a 24-hour average is required. The particle indices are defined on 24-hour averages. For real-time reporting the EPA uses the NowCast weighting, which is a different calculation from the one on this page.
- Mixing averaging periods for gases. Ozone uses the maximum daily 8-hour average, CO the maximum 8-hour, NO₂ and SO₂ the maximum 1-hour. Applying an annual mean to any of these tables is meaningless.
- Using an outdated PM2.5 table. The fine-particle breakpoints changed with the 2024 standard revision. A concentration of 10 µg/m³ is Moderate on the current table and Good on the older one.
- Converting gases between ppb and µg/m³ carelessly. The conversion depends on molar mass, temperature and pressure. Use a proper concentration conversion rather than a single fixed factor.
- Reading an AQI above 500 as a valid index value. The published tables end there. The EPA describes higher concentrations as beyond the AQI, and any number quoted past the end of the table comes from an extrapolation someone chose, not from the method.
Which document governs
The index, its breakpoints, the reporting requirements and the category descriptors are set out in 40 CFR Part 58, Appendix G, and elaborated in the EPA's technical assistance document for reporting the AQI. Breakpoints change when the underlying National Ambient Air Quality Standard for a pollutant changes, because AQI 100 is defined to coincide with the standard — which is exactly what happened to PM2.5 in the 2024 revision. If you are reproducing a historically reported value, use the table in force at the time. Other countries operate their own indices with different breakpoints, categories and even scale ranges; a Chinese or Indian AQI of 150 is not the same air as a US AQI of 150.
Key terms
- Sub-index
- The AQI value computed for one pollutant on its own. The overall AQI is the largest sub-index across the pollutants being reported.
- Driving pollutant
- Also called the responsible or critical pollutant: the one whose sub-index equals the overall AQI, and therefore the one the health advice should address.
- Breakpoint
- A concentration marking the edge of an AQI band. Each band is defined by a pair of concentration breakpoints and a pair of index values, and the index varies linearly between them.
- NowCast
- A weighted average of recent hourly particle measurements the EPA uses for real-time reporting, giving more weight to the most recent hours so that the index responds quickly when conditions change.
- NAAQS
- National Ambient Air Quality Standards — the health-based concentration limits the EPA sets for the criteria pollutants. Each pollutant's standard is anchored at AQI 100.
Where the AQI fits, and where it does not
The AQI exists to support a same-day decision: whether to run outside, whether to keep a school class indoors, whether a construction crew needs respiratory protection. It compresses a great deal into one integer for exactly that purpose, and it should not be pushed further than that. Long-term exposure assessment uses annual mean concentrations against the annual standard, not index values, because the health effects of chronic exposure follow the mean rather than the peaks.
Three limits are worth stating plainly. The index covers six pollutants and no others. It reports the maximum rather than combining pollutants, so it under-represents mixed exposure. And it is a US construction — indices published elsewhere use different breakpoints and category names, so international comparisons of index values are not meaningful without converting back to concentrations first.
Where the AQI is driven by particles, the source often traces back to combustion, and quantifying that source is a separate exercise: see the fuel combustion calculator for the emissions side of the same fire, and the driving emissions calculator for the transport contribution. On the exposure side, the sibling tools in this category deal with the other main environmental exposure that is measured on a logarithmic health scale — noise — through distance attenuation and occupational dose.