Ambient air pollution, especially fine particulate matter (PM2.5), has been widely documented as a major global health concern linked to premature mortality and various chronic conditions. However, key limitations persist in the literature, including sparse individual-level data, imprecise historical exposure assessments, and uncertain confounding mechanisms. The UK Biobank, with its large sample size (>500,000 participants), comprehensive set of covariates, and detailed residential histories, provides an ideal platform to overcome these limitations. This thesis aims to quantify the chronic health effects of PM2.5 using novel linkage frameworks, state-of-the-art epidemiological designs, and extensive confounder adjustment. First, a linkage framework was developed to assign daily PM2.5 exposures at a 1 × 1 km spatial resolution to each participant’s residential address, accounting for changes in address over time. Second, time-to-event analyses were conducted using Cox Proportional Hazards models with time-varying exposures to estimate associations between chronic PM2.5 exposure and all cause and cause-specific mortality, and cardiovascular hospital admissions. These models were further enhanced with distributed lag approaches to capture long-term lag structures. Finally, potential confounding pathways were examined using both individual-level (e.g., smoking, body mass index, physical activity) and area-level (e.g., neighborhood deprivation) data, facilitated by directed acyclic graph (DAG) analyses and stratified regression. The exposure-linkage process yielded high-resolution daily PM2.5 estimates for the entire UK Biobank cohort, revealing substantial spatiotemporal heterogeneity. The output of the survival models demonstrated that participants exposed to higher levels of PM2.5 experienced elevated risks for both all-cause and cardiovascular mortality, with effects persisting at relatively low pollutant concentrations. Analyses of hospital admissions underscored associations between PM2.5 and an array of cardiovascular outcomes, including stroke subtypes and myocardial infarction. Distributed lag analyses indicated that 1-year and 3-year exposures exerted significant impacts, while 5-year averages sometimes revealed larger cumulative risks. Additional exploration of confounding established that area-level covariates, in particular recruitment centre and deprivation as well as individual behaviors each contributed to partially attenuating but not eliminating PM2.5-related associations, highlighting a complex relationship between pollution levels and socio-behavioral factors. These findings extend prior evidence by pinpointing key methodological advances in long-term PM2.5 exposure assessment, demonstrating that temporally refined and spatially granular pollution models can uncover associations obscured in coarser exposures. The results also show that confounders at both the individual and neighborhood levels are relevant to unbiased risk estimation, reaffirming the necessity of multi-scale data integration. Notably, the analyses highlight that meaningful health impacts may occur below current UK national regulatory thresholds, raising important public health questions regarding existing air quality standards. This thesis demonstrates that integrating rich cohort data with high-resolution spatiotemporal air pollution estimates enables more accurate quantification of long-term PM2.5-related health risks. Results emphasize that carefully addressing exposure assignment and multifactorial confounding can substantially refine effect estimates. The observed associations persist at relatively low exposure levels, suggesting a need for continued regulatory efforts. Beyond producing robust estimates for the UK context, the frameworks described here can be applied to other cohorts and environmental exposures,ultimately informing global strategies to reduce pollution-related disease burdens