Heterogeneous effects from interventions often remain hidden in between-group analyses, risking overgeneralized conclusions of treatment effects. In this exploratory study, we performed latent profile analysis to unveil differential treatment effects among children in The Incredible Years Teacher Classroom Management Programme (IY TCMP). This program has previously been shown to reduce behavioral problems in preschools and schools in total samples and subgroups. A total of 726 children (48.7% girls; M[subscript age] = 4.21 years; SD[subscript age] = 0.86) from 92 childcare centers in Norway participated in either the intervention (n = 338) or the matched control condition (n = 388). First, by conducting latent profile analysis on baseline levels of child-teacher relationship (closeness, conflict), behavioral problems, and social competence, three distinct profiles were identified: "High Risk" (26.4%), "Moderate Risk" (42.8%), and "Low Risk" (30.7%) -- each profile with unique characteristics. Second, we tested for within-profile, condition-by-time interactions following the intervention, showing distinct treatment responses for each profile. High-risk profiles profited most from the IY TCMP, with a substantial decrease in externalizing problems, more teacher closeness, and less teacher conflict. Moderate-risk profiles also gained better teacher-child relationships and improved social competence. The Low-Risk profiles showed no effects from the intervention. It is argued that latent profile analysis presents a feasible approach for examining within-sample heterogeneity in intervention research. It also reveals crucial information on treatment variability, as demonstrated in the Incredible Years Programme.