We predict effects of behavior patterns and status on performance expectations and group inequality using an integrated theory developed by Fisek, Berger and Norman (1991). We next test those predictions using new experimental techniques we developed to control behavior patterns as independent variables. In a 10-condition experiment, predictions accounted for about 72 percent of variance in the data, with closer fit for women than for men. The gender difference may be related to features of the experiment, especially to the experimental design that counters cultural gender prescriptions in some conditions. We suggest ways to improve the experiment by more precisely separating behavior from inferred performance competence in later research. Applications include using behavior to reduce undesirable effects of status generalization. (Contains 4 tables, 5 figures, and 19 notes.)