An evaluation was made of four Head Start bilingual/ multicultural curriculum models to assess their effectiveness and impact on children, staff, and parents. Intended as a pre-post design (with 90 children at each of eight Head Start replication sites and with treatment and control groups stratified on the basis of Spanish or English language preference, age, sex, and prior preschool experience), the study was conducted on an incompletely randomized sample of 243 experimental-group and 199 comparison-group children. Child competency measures assessed socioemotional development and Spanish and English language production, comprehension, and concept development. Parent interviews assessed attitudes and knowledge about education, expectations and aspirations regarding the child's educational achievement, and parental involvement in the child's education. Surveys of Head Start classroom staff focused on the staff's understanding of bicultural/bilingual children and their attitudes and sensitivity towards these children as well as their ability to appropriately modify instruction. Implementation forms, ethnographic notes, and focused observation of individual children were also used to gather data. Analyses of variance and covariance assessed differences in test performance of experimental and comparison children. Contrasts were made at the level of the overall study sample, at the level of the combined sites, and at the level of the individual sites. Results and their implications are discussed. (RH)