This study analyzed the critical features of three delivery modes for Head Start services in order to: (1) compare the effects of different delivery modes on the immediate outcomes for children and parents; and (2) explore the pattern of effects both within and across modes to ascertain how the process works. The first mode represented the most traditional mode of Head Start programming: a five-day per week, half-day, center-based program in which parent participation was encouraged and where home visits occurred at least three times per year. The second mode incorporated elements of center-based and home-based programming. Children participated in center-based activities two days per week and each family received one home visit each week. The third mode represented a home-based program where home visitors had responsibility for implementing all Head Start services. A total of 174 parent/child dyads participated in the study. Data were collected in two waves over two years. Four measures of learning opportunities were included, multiple sources of measurement were used to assess child outcomes, and four sources of information concerning the parents and the home were utilized in the study. Results suggest that: (1) it is easier to produce parent and home changes through program intervention than it is to produce child changes that are directly attributable to intervention; (2) child changes associated with program modality are marginal; and (3) parent and home changes that appear to be susceptible to intervention are also the best predictors of child posttest scores. Data is displayed in 29 tables, and 32 references are included. Appendixes provide the child measures, the parent measures, and a staff questionnaire for internal use. (PCB)