Satellite aerosol optical depth (AOD) products have been used to estimate surface PM 2.5 in different parts of the world. However, some revealed good but some relatively poorer relationship between AOD and PM 2.5 . The increasingly available lidar-based aerosol extinction profiles provide insights into the boundary layer as well as residual above it. Here we report a study in Taiwan using four-year (2006–2009) MPLNet data to characterize aerosol vertical distribution. We derived haze layer height (HLH) from MPLNet aerosol extinction profiles and classified profile differences by mean PBL extinction (MPE) and near-surface extinction (NSE). The former represents the mean extinction within boundary layer and the latter the closest extinction to surface. The comparison of MPE versus NSE leads to three distinct classifications of aerosol profiles to help interpret the relationship between AOD and PM 2.5. The approximation of normalizing AOD AERONET by HLH closely follows MPE in correlating with PM 2.5 (≥0.8 with respect to season or ≥0.85 with respect to profile classification). The correlation resulted from AOD MODIS /HLH is systematically lower than that derived by AOD AERONET /HLH. PM 2.5 values are overall better estimated by profile classification than those derived by season. Better performance of PM 2.5 is obtained with the approximation (i.e., normalizing AOD by HLH) than that using AOD only. The performance metrics used in quantifying the relationship reveal improvements in uncertainty by 2.9 μg m −3 (or 20%) with AOD AERONET /HLH and 2.3 μg m −3 (or 15%) with AOD MODIS /HLH in comparison to using AOD only.