The evaluation of a behavioral screening procedure for the detection of both conduct disordered and withdrawn children was implemented in two phases among children enrolled in a rural Head Start program. The first phase focused on preliminary identification, through teacher rankings and classroom observation, of a population of the most withdrawn and the most conduct disordered children within individual preschool classrooms. The second phase involved standardized classroom observations of identified children by trained observers. Different behaviors were targeted for each group of children, either conduct disordered or withdrawn. Children who exhibited the highest frequencies of maladaptive behaviors were selected to participate in a group social skills intervention program. Three groups were selected, two conduct disordered and one withdrawn. Each group consisted of five children. A control group of non-referred children from the same classroom was also established. Validation of the screening procedure was carried out during the baseline phase of the social skills program by trained observers who were blind to the purpose of the study and to the nature of the groups. Thirteen sessions of observations of each group were taken during two 10-minute table tasks which involved coloring a picture. Behaviors observed included on-task, out-of-chair, and disruptive activities. Teachers' rankings proved to be the most important component of the screening process. Results, which suggest that the screening procedure successfully discriminated between withdrawn children and conduct disordered children as well as between these children and non-referred controls, are discussed. (Author/RH)