To better understand the evolutionary and ecological effects of dispersal, there is growing emphasis on the need to integrate direct data on movement behaviour into landscape-scale analyses. However, little is known about the general link between movement behaviour and large-scale patterns of dispersal and gene flow. Likewise, although recent studies suggest that nonrandom, directionally biased movement and dispersal can promote evolutionary divergence, the generality of this mechanism is unknown. We test the hypothesis that directionally biased movement and dispersal by plethodontid salamanders interact with the topography of headwater areas to affect genetic and phenotypic divergence. Movements by Gyrinophilus porphyriticus and Eurycea bislineata show contrasting directional biases: upstream bias in G. porphyriticus and downstream bias in E. bislineata. Consistent with predictions of how these biases interact with slope to affect dispersal and gene flow, genetic distance increased with slope in G. porphyriticus and decreased with slope in E. bislineata over a standardized distance of 1 km along six headwater streams. In both species, phenotypic divergence in relative trunk length was positively related to genetic divergence. These results indicate that landscape-scale patterns of dispersal and gene flow are closely related to movement behaviour in G. porphyriticus and E. bislineata, and underscore the value of information on movement behaviour for predicting and interpreting patterns of dispersal and gene flow in complex landscapes. This study also provides new evidence that directionally biased movement and dispersal can be important sources of intra- and interspecific variation in population divergence, and highlights the value of explicit, a priori predictions in landscape genetic studies.