Background: PM2.5 is associated with tuberculosis. However, the influence of its components on within-host Mycobacterium tuberculosis evolution remains unclear. We aimed to evaluate the associations of its components with population-level tuberculosis incidence and within-host pathogen evolutionary dynamics. Methods: We analyzed 417,112 eligible pulmonary tuberculosis cases across 95 county-level regions in Jiangsu, China (2011-2024) using quasi-Poisson generalized additive mixed models. Deep whole-genome sequencing of 2,602 pretreatment Mycobacterium tuberculosis isolates was performed to assess unfixed mutations (polymorphism rates, pN/pS ratios, functional enrichment). Effect estimates were reported as percentage changes per 10-μg/m³ pollutant increment. Findings: All components significantly elevated tuberculosis risk (p < 0.001). At a 6-month moving average, peak increases were 12.34% (95% CI: 11.95, 12.72) for PM2.5 951.36% (876.43, 1032.04) for black carbon, 129.11% (95% CI: 123.97, 134.37) for ammonium, and 83.34% (95% CI: 80.06, 86.67) for sulfate. We identified 162,328 unfixed mutations across 1,174 genes in 2,586 isolates. At 24-month exposure, PM2.5 and black carbon reduced nonsynonymous mutations (-18.00% [95% CI: -21.18, -14.69] and -99.02% [95% CI: -99.62, -97.49]) and pN/pS (-6.85% [95% CI: -9.17, -4.48] and -81.27% [95% CI: -89.72, -65.85]). Black carbon decreased pN/pS by 86.92% (95% CI: -93.92, -71.90) in positively selected genes at 13-months exposure, but increased it by 224.37% (95% CI: 45.82, 621.55) under purifying selection at 24-months exposure. Elevated PM2.5 constituent levels shifted differentially mutated genes from early DNA replication/repair to prolonged intermediary metabolism/respiration. Interpretation: PM2.5 components were associated with increased tuberculosis notifications and altered patterns of within-host unfixed genomic variation in Mycobacterium tuberculosis.