Human activity commonly has negative impacts on wildlife. Often, however, only a single element of the life cycle is affected, and it is unclear whether such effects translate into effects on population growth. This is particularly true for research into the causes of global amphibian declines, where experimental research focuses primarily on the aquatic larval stages but theory suggests these stages have only minor importance for population growth. We used data from long‐term mark‐recapture studies of two natural populations of the salamander Salamandra salamandra to confirm the predictions of population models. One population remained stable (i.e., stationary) throughout the 20 years of the study whereas the other declined to local extinction. We used mark‐recapture models to break down population growth rate into its two main components, recruitment and adult survival. Survival of postmetamorphic salamanders was constant over time in the stable population, whereas the declining population was characterized by a decrease in survival and constant recruitment. Population growth was most sensitive to variation in adult survival. Current amphibian research focuses on preadult stages, and researchers assume recruitment is the most important determinant of population growth. This may not be the case. A better understanding of amphibian population dynamics is possible only through the integration of experiments, theory, and data from natural populations. Our results also suggest that amphibian conservation efforts should focus on all stages of the life cycle and their associated habitats.