Serum adiponectin levels are decreased in obese subjects. We examined the association of current body weight (BW) and its change with a change in serum adiponectin levels. Serum adiponectin levels at the baseline (from 1995 to 1997) and the 5-year follow-up (from 2000 to 2002) examinations were evaluated in 1003 (M/F, 425/578; age at the baseline examinations, 58.3 +/- 11.7/57.5 +/- 11.0 years) Japanese subjects from a cohort population (N = 2013) of the Funagata study. Correlations and associations of BW at the baseline examinations and changes in BW between the baseline and the follow-up examinations (deltaBW) with changes in the serum adiponectin levels in the study period (deltaAdiponectin) were examined. Stepwise regression analyses revealed a significant correlation of the deltaBW (r = -0.233 and -0.204 for men and women, respectively; r = -0.324 for the upper tertile group divided based on their body mass index in women) with the deltaAdiponectin. However, the BW at the baseline examinations was not significantly correlated in both sexes. Multiple logistic regression analyses revealed that subjects who reduced their BW by 2 kg or more were 2.56 (95% confidence interval, 1.21-5.42; P = .014) and 8.24 times (95% confidence interval, 3.59-18.9; P < .001) more likely to be in the upper tertile of the deltaAdiponectin than those who increased their BW by 2 kg or more in men and women, respectively, independent of their BW at the baseline examinations. In conclusion, we showed here that the deltaBW was strongly associated with the deltaAdiponectin in both sexes, whereas the BW at the baseline examinations was not associated with the deltaAdiponectin, at least in women.