Objectives: Mental health problems (MHP) among medical residents are often found in clinical settings and sometimes lead to professional lapses. Evidence suggests that excessive Internet use is associated with MHP. We investigated the MHP of residents (depression, anxiety, and self-esteem) and their association with Internet use using a longitudinal design. Methods: Participants were 208 residents. The General Health Questionnaire (GHQ), Patient Health Questionnaire (PHQ), and the Rosenberg Self-Esteem Scale (RSES) were used to assess anxiety, depression, and self-esteem. The Generalized Problematic Internet Use Scale 2 (GPIUS2) was used to measure Internet use. Data were obtained twice, at baseline and 3 months later when the risk of MHP in residency is highest. Results: Residents with MHP (N = 36) had higher GHQ scores than those without MHP (N = 172) at follow-up. Residents with MHP had more depression and less self-esteem than those without MHP at baseline and follow-up. GPIUS2 total scores, and scores on the subscale preference for online social communication, were higher in residents with MHP. Preference for online social communication at follow-up was positively correlated with depression at baseline and follow-up, and negatively correlated with self-esteem at follow-up. Conclusions: Depression and self-esteem may predict Internet use when the risk of MHP is greatest for residents, indicating potential risks of excessive Internet use or Internet use as a compensatory coping behavior. Together with depression and self-esteem assessment, Internet use may be a useful index of resident mental health.