Objective: To compare body mass index (BMI), waist-hip ratio (WHR) and waist circumference as predictors of all-cause mortality among the elderly.

Design: Population-based cohort study; mean follow-up was 5.4 y.

Setting: The Rotterdam Study.

Participants: A total of 6296 men and women; baseline age 55-102 y.

Measurements: Sex-specific all-cause mortality was compared between quintiles of BMI, WHR and waist circumference and between predefined categories of BMI and waist circumference, stratified for smoking category.

Results: High quintiles of waist circumference, but not high quintiles of BMI and WHR were related to increased mortality among never smoking men, without reaching statistical significance. Only the highest category of BMI (BMI>30 kg/m2) among never smoking men was related to increased mortality, compared to normal BMI (hazard ratio 2.6 (95% confidence interval: 1.3-5.3)). Waist circumference between 94 and 102 cm and waist circumference 102 cm and larger were related to increased mortality, compared to normal waist circumference (hazard ratios 1.7 (95% confidence interval 1.1-2.8) and 1.6 (95% confidence interval 1.0-2.8), respectively). The proportion of mortality attributable to large waist circumference among never smoking men was three-fold the proportion attributable to high BMI. Among never smoking women and ex- and current smokers, categories of large body fatness did not predict increased mortality.

Conclusion: Among never smoking elderly men waist circumference may have more potential for detecting overweight than the BMI.