Delhi's PM2.5 Averaged 97 Micrograms per Cubic Metre in 2025, Down From 114 in 2018, but November Still Averaged 215 and the Source Studies Disagree on What Causes It

Delhi’s annual average PM2.5 in 2025 was 97 micrograms per cubic metre (µg/m³) and its PM10 was 198, according to a release from the Commission for Air Quality Management (CAQM) that the Press Information Bureau carried on 2 January 2026. PM2.5 and PM10 are particles smaller than 2.5 and 10 microns across. The limits the Central Pollution Control Board notified on 18 November 2009 are 40 and 60 µg/m³ as annual averages, so by tuput’s arithmetic Delhi’s 2025 PM2.5 was about 2.4 times its limit and its PM10 about 3.3 times.

Both figures are the lowest in CAQM’s 2018 to 2025 series apart from the Covid year 2020. The release credits “persistent field level efforts and targeted policy initiatives” for the improvement.

Eight years of annual averages, and what 2020 did

CAQM’s series starts in 2018. The 2 January release prints daily-average PM10 and PM2.5 for each calendar year, and a release of 31 December 2025 adds the air quality index (AQI) figures. The AQI is an index, not a concentration. CAQM counts 0 to 100 as “Good” or “Satisfactory”, 201 to 400 as “Poor” or “Very Poor”, and above 400 as “Severe”.

PM2.5 fell from 114 to 97 between 2018 and 2025, about 15 percent, and PM10 from 242 to 198, about 18 percent. The Covid year is the low point at 95 and 181. From 2021 to 2024 PM2.5 stayed between 98 and 105, and 2025 came in 7.6 percent under 2024.

Published versions of 2025 differ. The 31 December release, using data to 29 December, gave 96 and 197. The 2 January release, covering the full year, gave 97 and 198. The Delhi Economic Survey 2025-26 uses the Delhi Pollution Control Committee’s own monitors and gives 100 and 209 for 2025, and 110 and 225 for 2024. PTI reported that the survey’s 2025 figure used data to 9 December, which leaves out the last three weeks of a month that CAQM’s table averages at an AQI of 351. The survey’s series also falls from 2024 to 2025, so the two networks agree on direction and differ on level.

November is not July

The annual number averages two different Delhis. In CAQM’s monthly table the 2025 average AQI was 78 in July and 89 in August, then 354 in November and 351 in December. In every year from 2018 to 2025 the lowest monthly figure falls in July, August or September, the rainy months explained in our piece on how the monsoon works.

Concentrations take the same shape. The Centre for Research on Energy and Clean Air (CREA), reading the Central Pollution Control Board’s continuous monitors, put Delhi’s PM2.5 at 107 µg/m³ for October 2025 and 215 for November, with 23 “Very Poor” days and six “Severe” ones. November 2024 averaged 249, after 111 in October 2024. The Delhi Economic Survey, as PTI quoted it, records a PM10 reading of 1,154 µg/m³ on 15 May 2025 in a dust storm, and a PM2.5 peak of 451 on 21 October 2025, after Deepawali.

The 2018 study by the Automotive Research Association of India (ARAI) and The Energy and Resources Institute (TERI) found winter concentrations higher because lower wind speed and a shallower layer of mixing air reduced the atmosphere’s ability to disperse emissions. CAQM’s 31 December release blames still air and unfavourable weather for December 2025’s average AQI of 351.

Four studies, four rankings

Mukesh Sharma and Onkar Dikshit of IIT Kanpur submitted their study for the Delhi government in January 2016, on sampling at six sites from November 2013 to June 2014. Its emissions inventory, a count of what is released over a year, put road dust first for PM2.5 at 38 percent of 59 tonnes a day, then vehicles at 20 percent, household fuel at 12 and industrial point sources at 11. Its receptor model works backwards from the chemistry of particles caught on filters, and it gave a different winter picture for PM2.5: secondary particles, which form in the air from gases such as nitrogen oxides, at 30 percent, vehicles 25, biomass burning 26 and burning of municipal waste 8, with soil and road dust a smaller share. For summer the same model gave coal and fly ash 26 percent and soil and road dust 27.

ARAI and TERI, paid by the Department of Heavy Industry, reported in August 2018 on 20 sites across Delhi and the wider National Capital Region sampled from April 2016 to February 2017. Average PM2.5 across the sites was 90 µg/m³ in the summer sampling and 168 in winter. Their dispersion model, which simulates how emissions spread, gave Delhi’s winter PM2.5 as industry 30 percent, transport 28, dust 17 and biomass burning 14. Summer was dust 38, industry 22 and transport 17. The authors wrote that the model’s industry share “seems to be overestimated”, and that sampling skipped October, the month when farm burning usually peaks.

In March 2019 the Council on Energy, Environment and Water (CEEW) lined up five inventories from 2007 to 2018. Across the four that cover PM2.5, transport’s share ran from 17.9 to 39.2 percent and road dust’s from 18.1 to 37.8. Industry’s share was 28.9 percent in Guttikunda’s 2018 inventory and 22.4 in SAFAR’s, but 3.4 in TERI’s and 2.3 in IIT Kanpur’s. CEEW traced part of the gap to each study’s area, year, season and method, and said those factors explain it only partly. Each inventory covered a single year.

The newer tool is the Decision Support System (DSS) run by the Indian Institute of Tropical Meteorology (IITM) in Pune. Its authors’ 2023 preprint reports that for the post-monsoon and winter seasons of 2021-22, Delhi’s own emissions supplied 34.4 and 33.4 percent of the city’s PM2.5, other NCR districts 31 and 40.2, and biomass burning 7.3 and 0.1.

One third local, and the farm fire question

One finding repeats. About a third of Delhi’s PM2.5 comes from Delhi. TERI and ARAI put the city’s own share at 36 percent in winter and 26 in summer, the DSS preprint at 33.4 percent for winter, and the Centre for Science and Environment (CSE), reading the DSS for 1 to 15 December 2025, at about 35. In that reading vehicles made up 46 percent of Delhi’s local PM2.5, industry 22 and households 11.

They also agree that much of the fine particulate matter forms in the air. TERI and ARAI gave secondary particles 26 percent of winter PM2.5 and IIT Kanpur 30. CSE, using the DSS’s chemical data, reported 61 percent for 1 to 16 November 2025. The studies count different things in different years, and this article could not reconcile 61 with 26 to 30.

Farm fires show the widest spread. IIT Kanpur’s model gave all biomass burning, local and regional, 26 percent of winter PM2.5. The DSS preprint gave biomass-burning activities, which it tracks as fires in neighbouring states, 7.3 percent as a post-monsoon mean and 0.1 in winter. CREA, reading the same system, put stubble burning’s average share of November PM2.5 at about 20 percent in 2024 and 7 in 2025, with daily peaks of 38 and 22 percent.

What the policies left in the record

On CNG, the Supreme Court’s directions of July 1998 included converting the city bus fleet to compressed natural gas, Down To Earth’s timeline records. Urvashi Narain and Alan Krupnick of Resources for the Future analysed monthly data from 1990 to 2005 in a working paper of 27 February 2007. They found that converting buses cut PM10, carbon monoxide and sulphur dioxide. Converting three-wheelers did not give the same benefit, possibly because of poor technology, and a growing share of diesel cars and more vehicles offset part of the gain.

The Badarpur coal plant, rated at 720 MW, shut for good on 15 October 2018, The Wire reported, after the Environment Pollution (Prevention and Control) Authority recommended closure by mid-2018. IIT Kanpur’s 2013 to 2014 inventory had industrial point sources, mostly power plants, at 52 percent of Delhi’s nitrogen oxides and over 90 percent of its sulphur dioxide, and TERI and ARAI wrote that the city’s small sulphur dioxide emissions came mainly from Badarpur. Their winter model gave power stations 6 percent of Delhi’s PM2.5. None of the sources opened here measures the air before and after the closure.

On vehicle standards, Delhi received BS-VI fuel, with sulphur cut from 50 to 10 parts per million, on 1 April 2018, two years ahead of the rest of India, Hydrocarbon Engineering reported. On 24 October 2018 the Supreme Court ordered that no BS-IV vehicle be sold or registered anywhere in India from 1 April 2020. TERI and ARAI expected vehicle particulate emissions 49 percent lower in 2030 with BS-VI in place, and estimated that electrifying part of the fleet would take 5 to 6 percent off winter particulate concentrations in 2025 to 2030. National registrations are in our electric vehicle numbers piece.

The environment ministry notified the Graded Response Action Plan on 12 January 2017, and it took effect in Delhi-NCR that February, Down To Earth records. It adds restrictions as the AQI worsens. A modelling study by IITM and CAQM, reported on 8 January 2026, estimated that without the plan Delhi would have had around 60 percent more hours of “severe” air and 60 to 200 percent more of “severe plus”, with restrictions cutting PM2.5 by 2 to 42 µg/m³ depending on the stage. CAQM co-wrote it, and the news report names no journal. A study of nine monitored hotspots from September 2023 to March 2024, published in Discover Atmosphere on 15 April 2026, found PM2.5 stayed high at the higher stages and called the relief “limited and meteorology-dependent, short-term relief”.

What the averages cannot separate

The annual figure mixes weather with emissions. CSE found PM2.5 averaged 163 µg/m³ across 1 October to 30 November 2025, when farm fires were burning, and 210 from 1 to 28 December, when their share was about 0.2 percent. December’s PM2.5 was 24 percent above December 2024’s, in a release that has the full year about 7 percent lower than 2024.

November offers a rough comparison. Farm fires’ average share fell from about 20 percent to 7, and Delhi’s monthly PM2.5 fell from 249 to 215, or 14 percent. CREA counted 20 of 29 NCR cities with higher pollution than a year earlier. Its analyst Manoj Kumar said this indicates that year-round sources such as transport, industry and power plants are the dominant drivers.

None of the trend figures cited here is adjusted for weather, so the 15 percent fall since 2018 cannot be split between emission cuts and weather. The monitoring networks also disagree on levels, much as the regulator and the state board did over Prayagraj’s river water in our Ganga clean-up report.

CAQM’s release of 2 January 2026 says it expects “further gradual but marked improvement, year on year”. The same release puts 2025’s PM2.5 at 97 µg/m³, against an annual national limit of 40.