Aims To determine the effect of different glucose categories on incident cardiovascular disease (CVD) and all‐cause mortality in a population‐based cohort. Methods A total of 834 individuals aged 65 years and older without a history of CVD at baseline were stratified according to 2‐h post‐load glucose fasting glucose test into six categories including: (i) normal fasting glucose/normal glucose tolerance; (ii) prediabetes, (iii) isolated fasting hyperglycemia (IFH); (iv) isolated post‐challenge hyperglycemia (IPH); (v) IPH and IFH; and (vi) known diabetes mellitus. The prognostic significance of these groups on CVD and total mortality were examined by Cox proportional hazard ratios in a multivariate adjusted model. Results Over 9 years of follow up, 186 incidents of CVD and 218 deaths occurred (72 CVD mortality).Of the population, 45.2%, 30.7%, 1.2%, 6.1% 4.7%, and 11.9% were normal fasting glucose/normal glucose tolerance, prediabetes IFH, IPH, IFH and IPH, and known diabetes mellitus, respectively. Multivariate adjusted hazard ratios for CVD were 1.13 (95% CI 0.78–1.64), 1.03 (95% CI 0.25–4.22), 1.17 (95% CI 0.65–2.11), 2.52 (95% CI 1.43–4.42) and 2.39 (95% CI 1.55–3.69), and for CVD mortality were 0.59 (95% CI 0.27–1.30), 2.02 (95% CI 0.27–15.15), 1.26 (95% CI 0.51–3.16), 3.57 (95% CI 1.64–7.75), and 4.70 (95% CI 2.54–8.69) for prediabetes, IFH, IPH, IFH and IPH, and known diabetes mellitus phenotypes, respectively. Corresponding hazard ratios for all‐cause mortality in multivariate model adjusted for prevalent CVD were 1.07 (95% CI 0.73–1.57), 0.59 (95% CI 0.08–4.30), 0.92 (95% CI 0.5–1.70), 2.31 (95% CI 1.33–4.01) and 3.88 (95% CI 2.70–5.55), respectively. Conclusion Among the elderly population with newly diagnosed diabetes, only the combined IFH and IPH phenotype, but not IFH or IPH alone, was a significant predictor of CVD and mortality events. Prediabetes was not associated with any risk. Geriatr Gerontol Int 2016; 16: 1263–1271.