AbstractYoung adult men who gamble frequently face an elevated risk of developing gambling-related problems. So far, little is known about how changing levels of perceived social support interact with the course of gambling behaviour and gambling-related problems in this population. Using data from a prospective single-arm cohort study (Munich Leisure Time Study), we applied hierarchical linear models to investigate the longitudinal association of changes in perceived emotional and social support (hereafter PESS; operationalized as ENRICHD Social Support Instrument score) with gambling intensity, gambling frequency, and fulfilled criteria for gambling disorder. Pooling data from three time points (baseline, 12-month and 24-month follow-ups) to assess two 1-year intervals, these models disentangle the associations of (a) “level of PESS” (cross-sectional, between participants) and (b) “changes in individual PESS” (longitudinally, within-participants). Among the 169 study participants, higher levels of PESS were associated with fewer gambling-related problems (− 0.12 criteria met; p = 0.014). Furthermore, increasing individual PESS was associated with lower gambling frequency (− 0.25 gambling days; p = 0.060) and intensity (− 0.11 gambling hours; p = 0.006), and fewer gambling-related problems (− 0.19 problems; p < 0.001). The results suggest a mitigating influence of PESS on gambling behaviour and gambling-related problems. Increasing individual PESS appears more decisive for this pathway than high initial levels of PESS. Treatment and prevention strategies that activate and reinforce beneficial social resources in people with gambling-related problems are recommended and promising.