The reliability and meaning of assertiveness tests were explored using 120 female undergraduates. Several self-report inventories (the College Self-Expression Scale, Conflict Resolution Inventory, and a global rating from one to seven) were administered, as were three anxiety measures (Timed Behavior Checklist, response latency, and response duration) which were collected during administration of the College Women's Assertion Sample (CWAS). The CWAS involved 50 audiotaped interpersonal situations, during which the subjects' role-playing responses were observed. Five self-report measures of anxiety (the Fear of Negative Evaluation Scale, S-R Inventory of Anxiousness, Social Reaction Inventory, Autonomic Perception Questionnaire, and a global rating of anxiousness from one to seven) were then administered. Retesting at one-, four-, or ten-week intervals supported the reliability of the behavioral sampling and the Conflict Resolution Inventory. Convergent validity was evident for all the assertion tests, but only the CWAS showed strong discriminant validity. The global self-ratings were minimally useful, and the behavioral sampling successfully differentiated between aggressive, assertive, submissive, and aggressive-submissive subjects. The use of multiple measures was recommended. (Author/GDC)