Studies predicting service utilization of young children in community-based mental health programs are scarce. Linking Actions with Unmet Needs in Children’s Health (LAUNCH) was a pediatric mental health program to increases services for young children (0-8 years) at risk of socioemotional challenges in three primary care sites across three cities in Massachusetts. This study aims to identify baseline predictors of service utilization for families enrolled in LAUNCH. Data come from 114 families who consented to participate in the evaluation of LAUNCH. Service utilization was defined as the number of visits with the LAUNCH team. A priori, we identified covariates that might be associated with service utilization. Modified Poisson regression, adjusted for clustering at the site level, was used to model baseline covariates predicting service utilization. Further analysis is ongoing. Missing data will be accounted for using multiple imputation. Preliminary results suggest that in a multivariable Poisson model, caregiver depression scores and race significantly predicted service utilization. Increasing depression scores were associated with decreasing service utilization (Incident Rate Ratio (IRR)=0.97, p < .0001). African-American participants had a lower rate of service utilization compared to White participants (IRR=0.44, p < .05). Findings highlight the need to (1) address barriers to engagement for African-American families and (2) monitor and treat baseline caregiver depression to improve mental health service utilization for children at risk of socioemotional difficulties. Providers should consider the role of family-level factors when developing strategies to engage families of young children in pediatric mental health services.