Background: Access to university mental health services is poorly characterized. Our objectives were to (1) assess patterns of access and (2) explore predictability of contact with student mental health services. Participants: Data derived from the U-Flourish study, which includes a survey of successive cohorts of incoming undergraduate students attending Queen's University, located in Ontario, Canada (Cohort 1: 2018, Cohort 2: 2019). Methods: Survey data sets were deterministically linked to administrative data provided by Student Wellness Services. Analyses included cross-tabulation, logistic and negative binomial regression. Predictive modeling used LASSO regression. Results: Baseline symptoms were robust determinants of access. For example, a PHQ-9 rating in the severe range ([greater than or equal to] 20) was associated with an OR of 9.71 (95% CI: 4.46-21.1). A predictive algorithm did not outperform cut point-based interpretation of PHQ-9 or GAD-7 ratings. Conclusions: Self-reported symptoms are consistently associated with service use, supporting the widespread use of symptom screens.