Evaluation of Head Start has been based on four sources of information: (1) census surveys of children and families served and programs offered, (2) special research, projects on child development and experimental programs, (3) a longitudinal study of the development of low income children, and (4) a series of national evaluation studies. Available data appear to indicate that Head Start and other preschool programs have an immediate impact, but little is known about why, or under what circumstances optimum results may be obtained. Sustained gains are still being sought. Children who have not attended preschool programs tend to catch up in primary school with attenders, but little is known about why this happens. A planned variation study is in progress comparing children in sponsored Head Start and sponsored Follow-Through classes and children attending "regular" Head Start and "regular" primary schools. Head Start evaluations have tried to locate program variations other than administrative which may affect child development. Considering the evidence now available, the assumptions on which Head Start was based still seem tenable. Research is needed to clarify relationships between program and child variations, and the effects of long-term interventions. (NH)