Test estimates of intelligence and divergent thinking are compared with observations of social and nonsocial makebelieve play in a sample of 63 socially and economically disadvantaged preschool children. Controlling for the effects of age, indices of cognitive ability correlated positively and significantly with social make-believe play but not with solitary make-believe play. Uses Task performance scored for fanciful uses, but not for common uses, related to social fantasy play. Results suggest that social interaction is an important component in the relationship of fantasy play to cognitive development. A component of divergent thinking related to metaphorical or fantasy production appears to be a useful predictor of the propensity for sociodramatic play. (17pp.)