More carefully controlled research distinguishing more facets which may influence cognitive and social performance in young children is considered necessary before Head Start or any preschool program can be adequately assessed. Shortcomings in evaluation methodology discussed are as follows: evaluation research has not addressed itself to program input variables; preschool children are often evaluated as a homogeneous population; a disproportionate emphasis has been placed on cognitive development while ignoring the equally important and interrelated affective domain; studies of cognitive development have been too dependent on intelligence and achievement tests; and, preschool programs have been relied on too heavily as the only means of intervention. Wolf and Stein's (1966) follow-up research project on the effects of Head Start programs for Puerto-Rican, black, and non-Puerto-Rican white children in New York City is presented as a study which raises important issues in research methodology. Reviews of studies of cognitive development are presented, which support the theory that the degree of progress made by Head Start graduates and children without preschool experience can invariably be related to a child's base level of function. Problems of evaluating teacher influence and other variables of the affective domain are also discussed. (KG)