Air pollution is a major environmental and health problem, especially in urban agglomerations. Estimating personal exposure to fine particulate matter (PM2.5) remains a great challenge because it requires numerous point measurements to explain the daily spatial variation in pollutant levels. Furthermore, meteorological variables have considerable effects on the dispersion and distribution of pollutants, which also depends on spatio-temporal emission patterns. In this study we developed a hybrid interpolation technique that combined the inverse distance-weighted (IDW) method with Kriging with external drift (KED), and applied it to daily PM2.5 levels observed at 10 monitoring stations. This provided us with downscaled high-resolution maps of PM2.5 for the Island of Montreal. For the KED interpolation, we used spatio-temporal daily meteorological estimates and spatial covariates as land use and vegetation density. Different KED and IDW daily estimation models for the year 2010 were developed for each of the six synoptic weather classes. These clusters were developed using principal component analysis and unsupervised hierarchical classification. The results of the interpolation models were assessed with a leave-one-stationout cross-validation. The performance of the hybrid model was better than that of the KED or the IDW alone for all six synoptic weather classes (the daily estimate for R2 was 0.66–0.93 and for root mean square error (RMSE) 2.54–1.89 μg/m3).