Background and Aim: "Social norms" (SN)-interventions are aimed at changing existing misperceptions regarding peer substance use by providing feedback on actual norms, thereby affecting personal substance use. It is unknown whether SN-intervention effects previously demonstrated in US students can be replicated in German students. The aim of the INSIST-study was to examine the effects of a web-based SN-intervention on substance use. Design: Cluster-controlled trial. Setting: Eight Universities in Germany. Participants and Measurements: Students were recruited at four intervention vs. four delayed intervention control Universities. 4,463 students completed baseline, 1,255 students (59% female) completed both baseline and 5-months follow-up web-based surveys on personal and perceived peer substance use. Intervention participants received feedback contrasting personal and perceived peer use with previously assessed use and perceptions of same-sex, same-university peers. Intervention effects were assessed via multivariable mixed logistic regression models. Findings: Relative to controls, reception of SN-feedback was associated with higher odds for decreased alcohol use (OR: 1.91, 95% CI 1.42-2.56). This effect was most pronounced in students overestimating peer use at baseline and under or accurately estimating it at follow-up (OR: 6.28, 95% CI 2.00-19.8). The OR was 1.33 (95% CI 0.67-2.65) for decreased cannabis use in students at intervention Universities and was statistically significant at 1.70 (95% CI 1.13-2.55) when contrasting unchanged and decreased with increased use. Regarding tobacco use and episodes of drunkenness, no intervention effects were found. Conclusions: This study was the first cluster-controlled trial suggesting beneficial effects of web-based SN-intervention on alcohol and cannabis use in a large sample of German University students.