Objectives: To evaluate independent and interactive health effects of physical activity at work (PAW) and physical activity at habitual exercise (PAHE). Methods: A cross-sectional study on 1,117 male workers aged 29 to 46 years with a mean of 37.0 years. Using a self-administered questionnaire, the subjects were classified into `High'/`Low' (n=338/779) on PAW, and into `Yes'/`No' (n=353/764) on PAHE. As outcome indices, physical fitness expressed as maximal oxygen uptake (VO2max) was estimated, and serum total cholesterol (TC), HDL-cholesterol (HDLC) and TC/HDLC ratio were determined. Age, body mass index (kg/m2), alcohol-drinking and smoking were used as confounding factors. Results: The `High' PAW group had a significantly higher level of VO2max than the `Low' PAW group (+1.0 ml/kg min; 34.4 vs. 33.4 ml/kg min in adjusted means). The `High' PAW group also had a better profile of serum cholesterol than the `Low' PAW group, but it was not significant. The `Yes' PAHE group had significantly higher levels of VO2max (+2.0 ml/kg min; 34.9 vs. 32.9 ml/kg min) and serum HDLC (+0.09 mmol/l; 1.48 vs. 1.39 mmol/l), and a significantly lower TC/HDLC ratio (-0.29; 3.90 vs. 4.19) than the `No' PAHE group. Among PAW–PAHE subgroups, the `High–Yes' group (n=110), most physically active, had the best profile of VO2max and serum cholesterol, and the `Low–No' group (n=536), least physically active, had the worst one. Interactive effects of PAW and PAHE were not found either on VO2max or on serum cholesterol profile. All the results were not influenced by statistical adjustments for the four confounding factors used here. Conclusions: Both sedentary work and lack of habitual exercise are very common in Japanese workers. Sedentary work, independently from lack of habitual exercise, may increase the risk of diseases related to physical inactivity by affecting physical fitness and serum cholesterol profile.