Dozens of morphologically differentiated populations, subspecies and species of crossbills (genus Loxia) exist. It has been suggested that this divergence is due to variation in the conifer cones that each population specialises upon, requiring a specific beak size to efficiently separate the cone scales. If so, apparent survival should depend on beak size. To test this hypothesis, we undertook multievent capture–recapture modelling for 6844 individuals monitored during 27 years in a Pyrenean common crossbill L. curvirostra population in a forest of mountain pine Pinus uncinata. Apparent survival was indeed related to beak width, resulting in stabilizing selection around an optimum that was close to the observed mean beak width, indicating that local crossbill beak morphology is adapted to the conifer they feed upon. Both natural selection (selective mortality) and selective emigration of maladapted individuals may explain our findings. As is often the case in capture–recapture analyses but rarely recognised, we could not formally decompose apparent survival into selective mortality versus selective permanent emigration. Nonetheless, there are several indications that selective permanent emigration should not be fully excluded. First, natural selection by itself would have to be unusually strong compared to other empirical estimates to create the observed pattern of apparent survival. Second, the observed mean beak width was a bit lower than the estimated optimum beak width. This can be explained by immigration of crossbills with smaller beaks originating from southern populations, which may subsequently have left the study area permanently in response to low food intake. This is in line with a detected transient effect in the data, yet apparently little influx from crossbills from northern Europe. When permanent emigration is phenotypically selective this will have ecological and evolutionary consequences, so this possibility deserves more attention in general.