This study characterizes the structure and composition of mangrove forests across urban gradients in Puerto Rico. It then uses a suite of hydrologic, water chemistry, and land cover variables to test for the relative importance of urban intensity alongside flooding and water chemistry in explaining observed variability in forest structure and composition. Three separate statistical tests suggest a significant but limited influence of urbanness on forest composition and structure. In the most urban sites, the diameters of the largest trees were 27% larger, but all structural measurements were best explained by surface water chemistry, primarily nitrogen concentrations. Concentrations of ammonium and total Kjeldahl nitrogen best explained stem density, tree girth and canopy height. The most urban forests also contained 5.0 more species per hectare, on average, than the least urban forests, and simple regression suggests that urban metrics were the most powerful predictors of forest composition. The most urban forests were more dominated by Laguncularia racemosa, while both Avicennia germinans and Rhizophora mangle were found to be less abundant in the most urban sites, a trend that may be linked to the influence of precipitation and tidal connectivity on porewater salinity across the urban gradient. In multiple regression, no statistical difference was detected in the importance of surrounding land cover, flooding, or water quality in explaining the variance in either composition or structural metrics. This suggests that while a given forest metric may be strongly linked to either land cover, water quality, or flooding, all three are likely important and should be considered when characterizing these forests. With more human dependents in urban areas, the provisioning of important ecosystem services may be influenced by land use variables in addition to the more commonly measured metrics of water chemistry and flooding.