We assessed the major factors regulating adiponectin levels and the influence of exercise training on adiponectin levels in young obese men (19.2 +/- 1.1 yrs, BMI: 31.1 +/- 4.2, %fat: 27.2 +/- 3.9%). Subjects were separated into three groups (aerobic exercise group [AE: n=7], aerobic and resistance exercise group [AE + RE: n = 7], control group [n = 7]). AE underwent an 8-week training program (three times per week, more than 30 min endurance exercise at ventilatory threshold (VT) intensity). AE + RE went through resistance exercise two or three times per week together with the above endurance exercise for 5 months. Prior to intervention (n = 21), adiponectin levels were significantly correlated with percentage of fat. Stepwise multiple regression analysis revealed that percent body fat was an independent predictor of basal adiponectin levels (r2 = 0.370; P<0.01). After intervention, fat mass, and VT were significantly improved in AE. AE + RE exhibited significant reduction in weight, BMI, percent body fat and fat mass, and had significantly increased VT, VO2max cycling power and torque. Insulin was not changed in both groups. The control group exhibited no significant change in any variables. Although adiponectin levels were unchanged in the three groups, a significant negative correlation between delta fat mass and delta adiponectin levels was observed (n = 21, r = -0.461, P<0.05). In addition, delta percent body fat was an independent predictor of delta adiponectin levels (r2 = 0.327, P<0.05). These findings indicate that for increasing the adiponectin level, improvement of the body composition of young obese men is more important than the way training is performed.