Background: Creatinine-to-cystatin C ratio is recently suggested to be a surrogate marker for sarcopenia. However, little is known about its association with diabetes. This study aimed to fill in this gap based on a large-scale prospective cohort. Methods: A population-based representative sample of 5,055 participants aged ≥45 years from the China Health and Retirement Longitudinal Study was enrolled between 2011 and 2012 and followed at least once during the subsequent surveys at 2013, 2015, or 2018. Creatinine-to-cystatin C ratio was calculated and normalized by body weight. Incident diabetes was ascertained by plasma glucose, glycosylated hemoglobin, self-reported history, or use of anti-diabetic drugs. Logistic regression analysis and mediation analysis were employed. Results: During follow-up, 634 participants developed diabetes. The risk of diabetes was gradually and significantly decreased with increased normalized creatinine-cystatin C ratio. The multivariable-adjusted odds ratio for diabetes was 0.91 (95% confidence interval, 0.83 to 0.99) per 1 standard deviation higher of normalized creatinine-to-cystatin C ratio, and this relationship remained significant after controlling for muscle strength. The risk reduction in diabetes was significantly larger in participants with normal-weight and high normalized creatinine-to-cystatin C ratio compared with those with overweight/obesity and high normalized creatinine-to-cystatin C ratio (Pinteraction =0.01). Insulin resistance and inflammation appeared to be key mediators accounting for the observed relationship between normalized creatinine-to-cystatin C ratio and risk of diabetes, with their mediating effect being 93.1% and 22.0%, respectively. Conclusion: High normalized creatinine-to-cystatin C ratio is associated with reduced risk of diabetes in middle-aged and older adults.