Baltimore City Public Schools' school-based mental health program sought to reduce barriers to treatment for students by providing on-site mental health care. This study examines differences in treatment attendance, dropout, and pharmacologic referral among students enrolled in the Johns Hopkins affiliated program. Enrolled student records for SY2015-2016 and SY2016-2017 (N = 655) were analyzed. Latent class analyses (LCAs) identified classes of youth who varied in demographic and clinical characteristics, and differences in treatment outcomes by class were examined. LCAs identified three classes: one with higher probability of being male, Black, in elementary/middle school, and with externalizing symptoms (33%), one with higher probability of being female, Black, in high school, and with internalizing symptoms or adjustment problems (34%), and one with higher probability of being male, ethnically diverse, in elementary/middle school, and with equal probability of having or not having externalizing symptoms (33%). Classes differed significantly in their likelihood of receiving a psychiatrist visit during treatment (X[superscript 2] = 10.7, p = 0.01), with younger classes more likely to receive psychiatric visits than the older class. No class had fewer than 50% of students attending more than 20 sessions. However, classes differed in achieving high treatment attendance (X[superscript 2] = 9.6, p = 0.01), with both younger classes more likely to exceed 20 visits than the older class of students. The older class of students was more likely to refuse treatment once it was initiated (X[superscript 2] = 7.1, p = 0.03). Overall, the ESMH delivery-of-care model studied showed great promise in retaining vulnerable students in treatment, though older students are more likely to stop treatment and questions remain about the clinical benefit of so many students remaining in treatment for such a prolonged period of time.