This study is to prospectively investigate the association between serum uric acid and the incidence of type 2 diabetes in middle-aged and elderly Chinese. This study consisted of 924 non-diabetic adults aged 40 years or older at baseline. Subjects who received antidiabetic therapies and those who responded positively to the 75-g oral glucose tolerance test according to the 1999 World Health Organization criteria were diagnosed as having type 2 diabetes. Ninety-eight subjects developed type 2 diabetes during the 3.5-year follow-up. The hazard ratio (HR) for incident diabetes was 1.50 [95% confidence interval (CI) 1.18-1.92] for the highest sex-specific quartile of serum uric acid compared with the lowest after controlling for confounders. Participants with hyperuricemia had an HR of 1.95 (95% CI 1.11-3.44) for incident diabetes compared with those without hyperuricemia. Compared with the lowest quartile, the highest quartile had an HR for incident diabetes of 2.45 (95% CI 1.39-4.33) in men and 1.39 (95% CI 1.04-1.84) in women after fully adjustment. Adding serum uric acid to a model of conventional risk factors for diabetes improved the area under the curve for prediction of type 2 diabetes by 5%. Serum uric acid was an independent predictor of incident type 2 diabetes in middle-aged and elderly Chinese.