Abstract The world is rapidly urbanizing, which may influence physical activity (PA) levels - although little evidence is available for low- and middle-income countries. We evaluated associations between urbanization and total PA, as well as work-, leisure-, home-, and transport-specific PA, for 138,206 adults (35-70 years) living in 698 communities across 22 countries within the Prospective Urban and Rural Epidemiology (PURE) study. The 1-week total PA long-form International PA Questionnaire (IPAQ) was administered at baseline (2003-2015) and we used satellite-derived population density and impervious surface area to quantify levels of urbanization for 5 and 10 years prior to PA measurements. Generalized linear mixed effects models were used to examine associations between urbanization measures and PA, controlling for individual, household and community factors. Higher community baseline levels of population density (-12.4%, 95% CI: -16.0%, -8.7% per IQR) and impervious surface area (-29.2%, 95% CI: -37.5%, -19.7% per IQR), as well as 5-year population density change (-17.2%, 95% CI: -25.7%, -7.7 per IQR) was associated with lower total PA. Important differences in the associations between urbanization metrics and PA were observed between PA domains, country income levels, urban and rural status, and gender. These findings provide new information on the complex associations between urbanization and PA.