Background. Regular exercise reduces morbidity and mortality from many chronic diseases. However, in the year 2004 only 29% of Thai people exercise regularly and females (25.4%) exercised less than males (32.8%). Female nurses are believed to have similar roles and responsibilities to Thai women, therefore, served as study subjects. Purpose. This study proposed to test the model of exercise behavior of Thai female registered nurses in an urban hospital. Design. This descriptive study employs a cross-sectional design to examine the relationships among the variables in the model of exercise behavior which was developed based on Bandura's self-efficacy theory and the concepts of social support and physical environment. Methods. Siriraj Hospital located in Bangkok, Thailand, served as the study setting. Surveys were distributed to all female registered nurses potentially eligible to participate in the study. The original analysis plan was to employ structural equation modeling using the two-step approach. The measurement model was tested at the first step and demonstrated a poor fit; therefore, the full model was not performed at the second step. Path analysis was conducted to test the model of exercise behavior. Logistic regression was performed to test whether participants who exercised regularly and those who did not were correctly predicted by the proposed variables. Results. Self-efficacy for exercise had direct and indirect effects on exercise through outcome expectations for exercise. Family and friend social support had direct and indirect effects on exercise through self-efficacy and outcome expectations for exercise. Convenient facilities had a direct effect on exercise and presence of dependent person(s) had a direct effect on self-efficacy for exercise. Although the model did not fit the data it explained 12% of variability in exercise behavior. Logistic regression revealed acceptable fit. Self-efficacy for exercise, outcome expectations for exercise, family and friend social support, and age were significant predictors of exercise status. Conclusion/Implications. The supported relationships in the model of exercise behavior provided directions for the intervention development to increase regular exercise among female nurses. The findings indicate the need to reinforce the nurses on the significance of personal health care as it is their obligation to be a health role model to the public.