Physical activity is associated with several benefits, including maintaining a healthy weight, yet most American adults are not physically active (USDHHS, 2000). Futhermore, women and racial minorities are at a greater risk of being sedentary and overweight (USDHHS, 2000). A popular theory used to examine exercise determinants is the theory of planned behavior (TPB; Ajzen, 1985, 1988, 1991). The purpose of this research was to compare the TPB relationships between two different racial groups. Participants were 115 overweight women (M age = 32.35, SD = 12.66) enrolled in a 4-week exercise and diet program. Racial groups were categorized as Caucasian (N = 76) and nonCaucasian (N = 39). The nonCaucasian group included individuals who identified themselves as African American or black (n = 17), American Indian (n = 1), Asian or Pacific Islander (n = 7), Hispanic or Latino (n = 12), multiracial (n = 1), and other (n = 1). The program included three group exercise classes per week and a 1500-calorie diet. Participants rated their attitude, subjective norm, perceived behavioral control, and intention to attend the group exercise classes before beginning the program. Exercise behavior was defined as attendance of the group exercise classes. Results of the hierarchical regression analyses revealed that attitude, subjective norm, and perceived behavioral control explained 54.8% of the variance in intention for the Caucasian women, F (3, 68) = 27.520, p < .001, and 79.3% of the variance in intention for the nonCaucasian women, F (3, 35) = 40.828, p < .001. Perceived behavioral control was the only significant determinant of intention for both groups, however it was a stronger predictor of intention for the nonCaucasian women (b = .810, p < .001) than for the Caucasian women (b = .749, p < .001). For exercise behavior, intention alone (b = .266, p = .022) explained 7.1% of the variance for the Caucasian group, F (1, 72) = 5.475, p = .022. In comparison, neither intention nor perceived behavioral control was able to predict exercise behavior for the nonCaucasian group, F (2, 34) = 0.461, p = .635. The TPB appears to be sensitive to racial differences for the prediction of exercise intention, but may not be a reliable predictor of actual exercise behavior for overweight nonCaucasian women. These results have possible implications for the development of exercise interventions with special populations.Keyword(s): health promotion, physical activity, research