Purpose This study aimed to confirm the clinical impact of living arrangements on incidence of frequent alcohol consumption in university students. Design A retrospective cohort study. Setting A national university in Japan. Subjects 17,774 university students. Measures The association between living arrangements on admission and the incidence of frequent alcohol consumption (≥4 days/week) was assessed using multivariable-adjusted Cox proportional-hazards models. Results Among 5,685, 692, and 5,151 male students living with family, living in the dormitory, and living alone, 5.0%, 6.2%, and 5.8% reported frequent alcohol consumption during the median observational period of 3.0 years, respectively. Living in the dormitory and living alone were identified as significant predictors of frequent alcohol consumption (multivariable-adjusted hazard ratios: 1.00 [reference], 1.39 [1.01-1.92], and 1.21 [1.03-1.42], respectively). On the contrary, living arrangements were not associated with the incidence of frequent alcohol consumption among of 6,091 female students, partly because of low incidence of frequent alcohol consumption (2.3%, 1.4%, and 2.6%, respectively). Conclusions Living arrangements predicted frequent alcohol consumption among male university students, whereas not among female university students.