Different environments should select for different aspects of organismal performance, which should lead to correlated divergence in morphological traits that influence performance. The result should be genetic divergence in aspects of performance, morphology and associations (‘maps’) between morphology and performance. Testing this hypothesis requires quantifying performance and morphology in multiple populations after controlling for environmental differences, but this is rarely attempted. We used a common-garden experiment to examine morphology and several aspects of swimming performance within and between the lake and inlet populations of threespine stickleback (Gasterosteus aculeatus) from the Misty system, Vancouver Island, Canada. Controlling for body size, lake stickleback had shallower bodies, larger caudal fins and smaller pelvic girdles. With or without morphological covariates, lake stickleback showed greater performance in both sustained and burst swimming. In contrast, inlet stickleback showed greater manoeuverability than did lake stickleback in some analyses. Morphology–performance relationships were decoupled when considering variation within vs. between populations. Moreover, morphology–performance mapping differed between the two populations. Based on these observations, we advance a hypothesis for why populations adapting to different environments should show adaptive genetic divergence in morphology–performance mapping.