Assessment of youth psychopathology is frequently multi-informant, as each informant is presumed to provide unique information. However, clinicians frequently face a difficult task of integrating varying reports between informants as discrepancies in reports are common, particularly for less observable symptoms. Informant discrepancies have been linked to poorer treatment engagement and may contribute to the high rates of attrition in community-based services. The current study examined informant agreement on depressive symptoms among youths, parents, and clinicians to: 1) characterize patterns of agreement; 2) examine demographic and clinical predictors of agreement; and 3) examine the relationship between agreement and treatment engagement. It was hypothesized that levels of informant discrepancy would be high, predicted by sociodemographic and clinical factors, and predictive of poorer treatment engagement. The sample consisted of 326 youths (ages 13-18) with clinically elevated depressive symptoms (as reported by the youth, parent, or clinician) seeking services from an outpatient mental health clinic serving an ethnically diverse, low-income population. The analyses examined parent-youth-clinician (triadic) agreement, in addition to three sub-samples of dyads (i.e., parent-youth, parent-clinician, and youth-clinician). The study aims were examined using regression models (linear, logistic, and negative binomial) nested within clinician. Rates of dyadic agreement that the youth had clinically significant depressive symptoms ranged from 30.1% to 37.8%, while triadic agreement was 16.7%. Informant agreement was predicted by variables such as higher youth anxiety, higher perceived stress, and demographic factors (older age, female gender, and ethnic/racial minority status). A greater number of attended sessions was predicted primarily by agreement and clinical variables (lower youth anxiety). Greater consistency in treatment attendance was primarily predicted by agreement and demographic factors (being of non-minority status, not a recipient of public assistance). Clinician-defined dropout was not predicted by any of the variables of interest. Among adolescent youths with elevated depressive symptoms, agreement among informants is low. Informant agreement affects both how many sessions youths attend and consistency of attendance, particularly in the youth-clinician analyses. Increasing agreement on youth depressive symptoms at the initiation of treatment may have a positive effect on critical treatment processes, and in turn, treatment outcomes.