Objective This study investigated the association between long-term exposure to black smoke (BS) air pollution and mortality in two related Scottish cohorts with 25 years of follow-up.

Methods Risk factors were collected during 1970–1976 for 15331 and 6680 participants in the Renfrew/Paisley and Collaborative cohorts respectively. Exposure to BS during 1970–1979 was estimated by inverse-distance weighted averages of observed concentrations at monitoring sites and by two alternative spatial modelling approaches which included local air quality predictors (LAQP).

Results Consistent BS–mortality associations (per 10 μg m⁻³ increment in 10-year average BS) were observed in the Renfrew/Paisley cohort using LAQP-based exposure models (all-cause mortality HR 1.10 (95% CI 1.04 to 1.17); cardiovascular HR 1.11 (1.01 to 1.22); ischaemic heart disease HR 1.13 (1.02 to 1.25); respiratory HR 1.26 (1.02 to 1.28)). The associations were largely unaffected by additional adjustment for area-level deprivation category. A less consistent and generally implausible pattern of cause-specific BS–mortality associations was found for inverse-distance averaging of BS concentrations at nearby monitoring sites. BS–mortality associations in the Collaborative cohort were weaker and not statistically significant.

Conclusions The association between mortality and long-term exposure to BS observed in the Renfrew/Paisley cohort is consistent with hypotheses of how air pollution may affect human health. The dissimilarity in pollution–mortality associations for different exposure models highlights the critical importance of reliable estimation of exposures on intraurban spatial scales to avoid potential misclassification bias.