We investigated the differential effectiveness of a parent management training program for children with externalizing problem behavior. The parent management training was tested using a within-subject control group design. 270 families with children aged 3-10 years were included and assessments were made at 3 months before treatment, immediately before treatment, immediately after treatment, and at 12 months follow-up. We analyzed data using growth mixture modeling. For attention problems, we obtained a two-subgroup solution. One subgroup started with high initial values indicating more severe impairment, and the other subgroup had low initial values. The more impaired group showed stronger treatment effects. For disruptive behavior problems, our analyses revealed a three-subgroup solution. Two subgroups exhibited high initial levels of disruptive behavior problems and the third subgroup had low initial levels. One of the more impaired groups exhibited a strong decrease in problem behavior during treatment, while the other two groups only showed moderate decreases. For attention problems and disruptive behavior problems, some of the most impaired children gained the most from parent training.