Abstract Background This study examined the variation in city-level amenable mortality, i.e. mortality due to conditions that can be mitigated in the presence of timely and effective healthcare, in 363 Latin American cities and measured associations between amenable-mortality rates and urban metrics. Methods We used death records from 363 cities with populations of >100 000 people in nine Latin American countries from 2010 to 2016. We calculated sex-specific age-adjusted amenable-mortality rates per 100 000. We fitted multilevel linear models with cities nested within countries and estimated associations between amenable mortality and urban metrics, including population size and growth, fragmentation of urban development and socio-economic status. Results Cities in Mexico, Colombia and Brazil had the highest rates of amenable mortality. Overall, >70% of the variability in amenable mortality was due to between-country heterogeneity. But for preventable amenable mortality, those for which the healthcare system can prevent new cases, most of the variability in rates occurred between cities within countries. Population size and fragmentation of urban development were associated with amenable mortality. Higher fragmentation of urban development was associated with lower amenable mortality in small cities and higher amenable mortality in large cities. Population growth and higher city-level socio-economic status were associated with lower amenable mortality. Conclusions Most of the variability in amenable mortality in Latin American cities was due to between-county heterogeneity. However, urban metrics such as population size and growth, fragmentation of urban development and city-level socio-economic status may have a role in the distribution of amenable mortality across cities within countries.