This paper supports the hypothesis that picture vocabulary tests should not be used as interchangeable measures of intelligence for complex, lengthy intelligence tests (WISC and Stanford-Binet). In picture vocabulary tests assessing receptive functions (input), the child recognizes a word by pointing to or stating the number standing for an object. In WISC and Stanford-Binet tests assessing expressive functions (output), the child articulates the definition of a word. Subjects were drawn from three groups: (1) 56 retarded children from a psychodiagnostic laboratory; (2) 41 children from a child guidance clinic; and (3) 41 from multiply handicapped children. They were tested to see if differences in vocabulary input and output scores would vary according to physical setting, diagnosis, intellectual functioning, and educational achievement. The results showed that input-output differences may be predictive of educational achievement in children of borderline mentality but tend to be influenced by diagnosis, cerebral dysfunction or behavioral disturbance in children of average intelligence. Children's intellectual and educational potential is more precisely evaluated by judging both the input and output of a child's vocabulary. Correlations between picture and WISC vocabulary scores do not justify equivalence of measure and functions, especially with deviant groups of children. (DO)