Smoking causes various health problems. Limited studies have reported a clinical effect of skipping breakfast on smoking initiation among adolescents. This retrospective cohort study aimed to assess the dose-dependent association between skipping breakfast and smoking initiation in university students. This study included 17,493 male and 8880 female students aged 18−22 years at a national university in Japan. The association between breakfast frequency (eating every day and skipping occasionally, often, and usually) and smoking initiation was evaluated using Cox proportional hazards models adjusted for clinically relevant factors. Smoking initiation was observed in 2027 (11.6%) male and 197 (2.2%) female students over the median observational period of 3.0 and 3.1 years. Skipping breakfast was significantly associated with smoking initiation in a dose-dependent fashion in male students (the adjusted hazard ratios [95% confidence interval] of eating breakfast every day and skipping occasionally, often, and usually: 1.00 [reference], 1.30 [1.15, 1.46], 1.47 [1.21, 1.79], and 1.77 [1.40, 2.25], respectively). Female students skipping breakfast occasionally and often were more vulnerable to smoking initiation than those who ate breakfast every day (1.00 [reference], 1.86 [1.24, 2.78], 2.97 [1.66, 5.32], and 1.76 [0.55, 5.64], respectively). Breakfast frequency may be useful to identify university students at risk of smoking initiation who need improvement in their health literacy.