Background Growing epidemiological evidence suggests an adverse relationship between exposure to air pollutants and cognitive decline. However, there is still some heterogeneity in the findings, with inconsistent results depending on the pollutant and cognitive domain considered. We aimed to evaluate if air pollution was associated with global and domain‐specific cognitive decline. Methods We used data from the French Three‐City prospective cohort on 6380 participants aged 65 and older at recruitment and followed up for 12 years. Participants did cognitive tests at baseline and every 2 years, to assess global cognition (Mini Mental State Examination, MMSE), visual memory (Benton Visual Retention Test, BVRT), semantic fluency (Isaacs Set Test, IST) and executive functions (Trail Making Tests A and B, TMTA/B). Land‐use regression models were used to estimate fine particulate matter (PM2.5), nitrogen dioxide (NO2) and black carbon (BC) exposure at participants’ residential address over the 5‐year preceding the baseline visit. Linear mixed models were performed for each pollutant to evaluate their association with cognitive decline. Results Participants median age was 73.4 (IQR 8.0), and 61.5% were women. Baseline median scores were 28 (3) for MMSE, 12 (3) for BVRT, 48.0 (14.0) for IST, 0.5 (0.2) for TMTA, and 0.2 (0.2) for TMTB. We found a significantly worse decline in global cognition among participants with higher PM2.5 exposure: for PM2.5 levels ranging from 19.6 to 30.4 µg/m3, an IQR increment in PM2.5 (1.5) was associated with greater MMSE decline (β: 0.003 0.0006;0.006 per year). Other associations were inconsistent in direction and of small magnitude. For instance, an IQR increment in PM2.5 was associated with a 0.03 (‐0.002;0.060) change in IST decline and a ‐0.0003 (‐0.0009;0.0003) change in TMTA decline. Conclusion We found an adverse association between PM2.5 exposure and decline in global cognition. No significant associations were observed for the other pollutants and cognitive domains.