Abstract Background After stressful event exposure, higher perceived social support is a well-established correlate of decreased risk for psychological symptoms, including depressive, anxiety and posttraumatic stress (PTS) symptoms. However, longitudinal data on the direction of this association and the stability of perceived social support are scarce and have yielded mixed results, with a particular lack of prospective studies. We aimed to investigate changes in perceived social support and bidirectional associations between perceived social support and psychological symptoms in a prospective, longitudinal study. Methods A sample of German soldiers was assessed before and after deployment to Afghanistan. Group-based trajectory modelling was used to investigate the stability of perceived social support and to identify possible distinguishable trajectories of perceived social support. Bidirectional associations between perceived social support (general and workplace) and psychological symptoms (depressive, anxiety and PTS) were examined using gamma regressions. Results Average levels of perceived general social support did not change, while perceived workplace social support increased slightly (t(344) = 5.51, p