Predictors of retention of participants in a longitudinal study and heterogeneity between communities were investigated using a multilevel logistic regression model. Data from the longitudinal outcome study of the national evaluation of the Comprehensive Community Mental Health Services for Children and Their Families program and information on communities participating in the evaluation were used to examine retention of participants at 6-, 12-, 18-, and 24-month follow-up interviews. Findings revealed that both child and family characteristics and community characteristics significantly influenced participation for up to 24-month follow-up. Specifically, behavioral and emotional problems at intake, referral source, proportion of total staff devoted to interviewers, and total amount of incentive paid out at baseline influenced retention of participants. In addition, significant random effects at both Level 2 and Level 3 indicated that retention varied significantly between communities and among children even after controlling for child and family characteristics.