<p>Mounting evidence links air pollution to dementia, the most prevalent cause of cognitive impairment in older people. Here we investigated individual and compound effects of particulate matters (PM₁₀, PM_2.5, PM_{<i>coarse</i>}, PM_<i>abs</i>) and nitric oxides (NO₂, NO) on risk of all-cause dementia, and its most common subtypes, Alzheimer’s disease (AD) and vascular dementia (VAD), using data from UK Biobank. We addressed factors that hinder causal interpretation of associations previously shown in the literature and their translation into clear public health policies. Specifically: 1) spatial confounding by area-level covariates, 2) collinearity among and identification of the most relevant air pollutants, and 3) the time window for pollution exposure. Furthermore, we used chronic obstructive pulmonary disease (COPD) and frequency of oily fish intake in positive and negative control analyses. We found NO₂ to be the strongest risk factor for dementia, especially when considering participants with longer permanence at residential address as proxy for longer periods (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\ge 5\)</EquationSource> </InlineEquation> years) of exposure (all-cause dementia hazard ratio HR=1.06, 1.02-1.11 per 9.86 <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\mu g/m^3\)</EquationSource> </InlineEquation> interquartile range). There was stronger evidence of an effect on risk for AD than VAD. Positive control analysis did not provide any evidence against causality, although the analyses of spatial confounding and negative control analyses revealed the presence of some residual bias, thus warranting care in the interpretation of the results. Together, our results highlight that targeting air pollution, in particular NO₂ levels, could inform preventive public health policies for dementia.</p>