Two basic problems are confronted in this research: (1) What is sex role differentiation?, and (2) How does one go about measuring it? Both questions revolve around construct validity and the adequacy of sex role measures. Most investigators have used dependent "objective-verbal" tests to measure sex role differentiation. These either involve questions about the differences between boys and girls, or the presentation of toys appropriate for one sex or another along with questions regarding their appropriateness. This study examines the construct validity of these "objective-verbal" tests. Comparative data on sex role differentiation in preschool children were collected from "objective-verbal" tests and naturalistic observations. One-hundred fifty-two 3-5 year-olds were administered the It-test, the Occupational Preference Test, the Nadleman Recall Test, the DeLucia Toy Preference Test, and the Draw-A-Person test. They were observed in their classrooms approximately once every two weeks, with observers noting instances of aggressive, dependent, and cooperative behavior. Child-child and teacher-child interactions were also observed. Findings showed dependent behaviors to be most frequent, with agressive behaviors next; cooperative behaviors were rare. When correlations were made results demonstrated that children who scored low, or less masculine on the sex role tests showed more dependent behaviors than children with high masculine scores. Evidence also strongly supported the belief that children with high masculine scores on the objective tests behaved more aggressively. (Author/MV)