The epidemiologic data supporting an association between physical activity and lipoprotein profile is controversial. We hypothesized that physical activity is positively associated with triglyceride levels and negatively associated with high density lipoprotein (HDL) cholesterol levels, and that these associations are mediated by insulin activity. We tested these hypotheses in 819 male and 1159 female healthy non-diabetic Mexicans, aged 35-64 years, residing in Mexico City. Physical activity was assessed using a modified Stanford 7-day Physical Activity Recall. Serum glucose and insulin were evaluated in the fasting state and 2 h following a glucose challenge. Crude analysis indicated that physical activity was not associated with HDL cholesterol in males, but was associated with HDL cholesterol in females (r = 0.06; p = 0.025). Physical activity was not associated with triglyceride concentration in males or females. Multivariate results showed that the positive association between HDL cholesterol and physical activity in women remained after controlling for age, estrogen status, body mass index, and smoking. Furthermore, there was a significant interaction between physical activity, HDL cholesterol and an index of insulin sensitivity. Women with higher levels of insulin sensitivity had higher levels of HDL cholesterol. However, among women at lower levels of insulin sensitivity, those with greater physical activity were significantly more likely to have higher HDL cholesterol values.