Conversion of forests to agriculture often fragments distributions of forest species and can disrupt gene flow. We examined effects of prevalent land uses on genetic connectivity of two amphibian species in northeastern Costa Rica. We incorporated data from field surveys and experiments to develop resistance surfaces that represent local mechanisms hypothesized to modify dispersal success of amphibians, such as habitat-specific predation and desiccation risk. Because time lags can exist between forest conversion and genetic responses, we evaluated landscape effects using land-cover data from different time periods. Populations of both species were structured at similar spatial scales but exhibited differing responses to landscape features. Litter frog population differentiation was significantly related to landscape resistances estimated from abundance and experiment data. Model support was highest for experiment-derived surfaces that represented responses to microclimate variation. Litter frog genetic variation was best explained by contemporary landscape configuration, indicating rapid population response to land-use change. Poison frog genetic structure was strongly associated with geographic isolation, which explained up to 45% of genetic variation, and long-standing barriers, such as rivers and mountains. However, there was also partial support for abundance- and microclimate response-derived resistances. Differences in species responses to landscape features may be explained by overriding effects of population size on patterns of differentiation for poison frogs, but not litter frogs. In addition, pastures are likely semi-permeable to poison frog gene flow because the species is known to use pastures when remnant vegetation is present, but litter frogs do not. Ongoing reforestation efforts will probably increase connectivity in the region by increasing tree cover and reducing area of pastures.