Objective: A quality-improvement study evaluated the feasibility of training mental health providers to provide mental health screening and relationship-based intervention to expand services for children 0 to 5 years of age in eight California county mental health systems from November 2002 to June 2003. State-level training was provided to more than 582 participants and county-level training to more than 5425 participants, including ongoing supervision. Method: Direct services and use of collateral services were tracked. Psychiatric symptoms were screened with new Mental Health Screening and Risk Assessment tools for 388 children (mean age, 34 months). At intake and after intervention (mean of 22 visits), an index sample (93 children) were further characterized by the Diagnostic Classification for Zero to Three and DSM-IV, and parent-child relationship was characterized by the Diagnostic Classification for Zero to Three Parent-Infant Relationship Global Assessment Scale. Providers reported that 41% of their service time was directed to the parent and child together, 35% to the parent alone, and 24% to the child alone. Results: The 93 index children and 295 children in a clinic reference sample were comparable, supporting generalizability. After intervention, Mental Health Screening and Risk Assessment scores were significantly lower. Global Assessment of Functioning scores improved (effect size, 0.35), as did the relationship (Parent-Infant Relationship Global Assessment Scale effect size, 0.16). Conclusion: Training mental health staff to provide treatment to infants and preschool children and families in public mental health settings is feasible and leads to an increase in numbers of children served. (Contains 5 tables.)