A quasi-disguised nonverbal attitude measure for young children was used to measure racial awareness and preference. Subjects were 429 first and third grade white and Negro public school children. Varied class activities were presented in 18 pen and ink sketches with different racial combination of teachers and children. These sketches were arranged in booklet form so that on any given page, three of the six racial compositions appeared, one in each of the three class activities. Children marked which class they would most prefer and which class they would like least. Choices that were random or based only on activities were eliminated. Analyses of variance nested hierarchical designs, with race, grade and sex as main factors and type of school attended (segregated or desegregated) as a nested factor within race. Results suggested that white subjects generally began the first grade with a clear preference for an all-white social surrounding in school and maintained that preference into the third grade. Negroes started the first grade with a slight preference for an all-white class, but by third grade clearly preferred Negro teachers and peers. White subjects' racial preferences were more pronounced than Negroes', and third graders showed clearer racial awareness than first graders. Integration had no measurable effect on racial preferences in the schools studied. (DR)