OBJECTIVE: To assess the interaction between leisure-time physical activity (LTPA) and occupational activity (OA) on the prevalence of obesity. DESIGN: Secondary data analysis of a population based cross-sectional US national sample (NHANES III). SUBJECTS: A total of 4889 disease-free, currently employed adults over age 20 y. MEASUREMENTS: Subjects body mass index (BMI) was categorized as (1) obese (BMI ! 30 kgam 2 ), or (2) non-obese (BMI`30 kgam 2 ). LTPA was divided into four categories: (1) no LTPA; (2) irregular LTPA; (3) regular moderate intensity LTPA; and (4) regular vigorous intensity LTPA. OA was grouped as (1) high OA and (2) low OA. Age, gender, race ± ethnicity, smoking status, urbanization classi®cation, alcohol consumption and income were statistically controlled. RESULTS: In all, 16.8% (s.e. 0.7) of the total subject population were obese (15.1% (s.e. 1.1) of men and 19.1% (s.e. 1.1) of women). Logistic regression revealed that compared to those who engage in no LTPA and have low levels of OA, the likelihood of being obese is 42% (95% CI 0.35, 0.96) lower for those who engage in no LTPA and have high OA, 48% (95% CI 0.32, 0.83) lower for those who have irregular LTPA and have high levels of OA, and about 50% lower for all those who have regular LTPA through moderate or vigorous activity levels regardless of OA level. CONCLUSION: When considering disease free adults above 20 y of age employed in high and low activity occupations, a high level of occupational activity is associated with a decreased likelihood of being obese.