BACKGROUND Visceral adiposity index (VAI) has recently been developed based on waist circumference, body mass index (BMI), triglycerides (TGs), and high-density lipoprotein cholesterol (HDL-C). We examined predictive performances for incident diabetes of the VAI per se and as compared to the metabolic syndrome (MetS) and waist-to-height-ratio (WHtR). METHODS Participants free of diabetes at baseline with at least one follow-up examination (5,964) were included for the current study. Weibull regression models were developed for interval-censored survival data. Absolute and relative integrated discriminatory improvement index (IDI) and cut-point-based and cut-point-free net reclassification improvement index (NRI) were used as measures of predictive ability for incident diabetes added by VAI, as compared to the MetS and WHtR. RESULTS The annual incidence rate of diabetes was 0.85 per 1000 person. Mean VAI was 3.06 (95%CIs 2.99-3.13). Diabetes risk factors levels increased in stepwise fashion across VAI quintiles. Risk gradient between the highest and lowest quintile of VAI was 4.5 (95%CIs 3.0-6.9). VAI significantly improved predictive ability of the MetS. The relative IDI and cut-point free NRI for predictive ability added to MetS by VAI were 30.3% (95%CIs 18.8-41.8%) and 30.7% (95%CIs 20.8-40.7%), respectively. WHtR, outperformed VAI with cut-point-free NRI of 24.6% (95%CIs 14.1-35.2%). CONCLUSIONS In conclusion, although VAI could be a prognostic tool for incident diabetes events, gathering information on its components (WC, BMI, TGs, and HDL-C) is unlikely to improve the prediction ability beyond what could be achieved by the simply assessable and commonly available information on WHtR.