OBJECTIVE:To elucidate the effect of long-term weight variability on C-reactive protein (CRP) levels. DESIGN: Cross-sectional study of the circulating CRP. SUBJECTS: A total of 637 Japanese men aged 40-49 y in1997. MEASUREMENTS: Serum CRP levels, body mass index in 1997 (current BMI), the slope of weight on age (weight-slope) representing an individual's weight trend of direction and magnitude, and the root mean square error around the slope of weight on age (weight-RMSE) representing the weight fluctuation magnitude, as calculated by a simple linear regression model in which each value of the subject's five actual weights (aged 20, 25, 30 y, five years ago, and current) was a dependent variable and the subject's ages independent variables. RESULTS: After adjustment for age and confounders, including smoking and health status, the odds ratios of elevated CRP (Z0.06 mg/dl) were 1.83 (95% CI: 1.25-2.69), 2.63 (1.69-4.11), and 10.31 (2.17-48.98) for upper normal-weight (BMI: 22-o25 kg/m 2 ), overweight (25-o30), and obese (Z30) persons, respectively, compared with lower normal-weight persons (18.5-o22). Adjusting for age, confounders, and current BMI, weight-slope was positively associated with CRP level especially among subjects with BMIZ25 kg/m 2 (trend Po0.01), and weight-RMSE was positively associated with CRP level particularly among subjects with BMI o25 kg/m 2 (trend Po0.05). CONCLUSION: Our results suggest a state of low-grade systemic inflammation not only in overweight and obese persons, but also in normal-weight persons with large weight fluctuation, possibly explaining in part the positive association between weight fluctuation and CVD.