Background: PM 2·5 air pollution is associated with increased risk of diabetes; however, a knowledge gap exists to further define and quantify the burden of diabetes attributable to PM 2·5 air pollution. Therefore, we aimed to define the relationship between PM 2·5 and diabetes. We also aimed to characterise an integrated exposure response function and to provide a quantitative estimate of the global and national burden of diabetes attributable to PM 2·5 . Methods: We did a longitudinal cohort study of the association of PM 2·5 with diabetes. We built a cohort of US veterans with no previous history of diabetes from various databases. Participants were followed up for a median of 8·5 years, we and used survival models to examine the association between PM 2·5 and the risk of diabetes. All models were adjusted for sociodemographic and health characteristics. We tested a positive outcome control (ie, risk of all-cause mortality), negative exposure control (ie, ambient air sodium concentrations), and a negative outcome control (ie, risk of lower limb fracture). Data for the models were reported as hazard ratios (HRs) and 95% CIs. Additionally, we reviewed studies of PM 2·5 and the risk of diabetes, and used the estimates to build a non-linear integrated exposure response function to characterise the relationship across all concentrations of PM 2·5 exposure. We included studies into the building of the integrated exposure response function if they scored at least a four on the Newcastle-Ottawa Quality Assessment Scale and were only included if the outcome was type 2 diabetes or all types of diabetes. Finally, we used the Global Burden of Disease study data and methodologies to estimate the attributable burden of disease (ABD) and disability-adjusted life-years (DALYs) of diabetes attributable to PM 2·5 air pollution globally and in 194 countries and territories. Findings: We examined the relationship of PM 2·5 and the risk of incident diabetes in a longitudinal cohort of 1 729 108 participants followed up for a median of 8·5 years (IQR 8·1–8·8). In adjusted models, a 10 μg/m 3 increase in PM 2·5 was associated with increased risk of diabetes (HR 1·15, 95% CI 1·08–1·22). PM 2·5 was associated with increased risk of death as the positive outcome control (HR 1·08, 95% CI 1·03–1·13), but not with lower limb fracture as the negative outcome control (1·00, 0·91–1·09). An IQR increase (0·045 μg/m 3 ) in ambient air sodium concentration as the negative exposure control exhibited no significant association with the risk of diabetes (HR 1·00, 95% CI 0·99–1·00). An integrated exposure response function showed that the risk of diabetes increased substantially above 2·4 μg/m 3 , and then exhibited a more moderate increase at concentrations above 10 μg/m 3 . Globally, ambient PM 2·5 contributed to about 3·2 million (95% uncertainty interval [UI] 2·2–3·8) incident cases of diabetes, about 8·2 million (95% UI 5·8–11·0) DALYs caused by diabetes, and 206 105 (95% UI 153 408–259 119) deaths from diabetes attributable to PM 2·5 exposure. The burden varied substantially among geographies and was more heavily skewed towards low-income and lower-to-middle-income countries. Interpretation: The global toll of diabetes attributable to PM 2·5 air pollution is significant. Reduction in exposure will yield substantial health benefits. Funding: US Department of Veterans Affairs.