To develop a method for quantifying risks of death and dementia in relation to vascular risk factors the Gothenburg H-70 1901–02 birth cohort was studied (n = 380, was followed over 20 years, with 103 incident dementia cases). Separate vascular risk factor indices were calculated using 23 vascular risk factors to predict: (i) dementia-free-survival, and (ii) incident dementia derived from post hoc optimal separation of affected and unaffected cases. Classification of adverse outcomes (dementia/non-dementia; alive/dead) was assessed using receiver–operator characteristic (ROC) curves, and the area under the curve (AUC). Each index showed high separation between affected and unaffected cases. For dementia/non-dementia, the AUC was 0.74 ± 0.02 for 10 year and 0.67 ± 0.02 for 20 year; for death/survival, the AUC was 0.75 ± 0.02 for 10 years and 0.79 ± 0.03 for 20 years. Of note, few items were important in both indexes, and most showed reciprocal effects (e.g. decreased the risk of death but increased the risk of dementia). Our results suggest that vascular risk factor indexes can give robust estimates of dementia and life span prognoses in elderly people, but death and dementia have different risk profiles. This may be because of death being a competing risk for incident late-onset dementia.