To clarify the associations between plasma adiponectin levels and ADIPOQ variants with type 2 diabetes incidence in a general Japanese population. We conducted a case-control study nested within the Japan Public Health Center-based Prospective Study. We measured plasma adiponectin levels and genotyped +45T>G (rs2241766) and +276G>T (rs1501299) in the ADIPOQ gene among 417 incident diabetes cases and 1197 control subjects matched by age, sex, and area. After potential confounding factor adjustment, the multivariable-adjusted diabetes odds ratios (ORs) were 0.59 (95% confidence interval [CI]: 0.51-0.68) and 0.68 (95% CI: 0.60-0.78) per 1 standard deviation increment in the log-transformed levels of total- and high-molecular-weight (HMW) adiponectin levels, respectively. However, the ADIPOQ variants were not significantly associated with plasma adiponectin levels (for total adiponectin, +45 P=0.15 and +276 P=0.08) and diabetes risk (+45 P=0.70 and +276 P=0.72) under the additive genetic model. Our prospective findings suggest that both total and HMW adiponectin levels are strongly and inversely associated with diabetes risk after adjustment for potential confounding factors; however, the ADIPOQ variants +45 and +276 are not associated with adiponectin levels and diabetes risk in the general Japanese population.