Quantification of the excess mortality from all causes associated with obesity remains controversial. In this paper, 6,193 obese patients, those with a body mass index (weight (kg)/height (m)2 (BMI)) range of 25-74 kg/m2, recruited from 1961 to 1994 in Düsseldorf, Germany, were followed for a mean time of 14 (standard deviation, 8.2) years, yielding 87,179 observed patient-years. During the study period, 1,028 patients (16.6%) died. The entire cohort was grouped into approximate quartiles according to BMI: group 1, BMI from 25 to < 32; group 2, BMI from 32 to < 36; group 3, BMI from 36 to < 40; group 4, BMI > or = 40 kg/m2. The following risk ratios were estimated by means of Cox proportional hazards models using the lowest BMI group as reference category: group 2 for men, 1.02 (95% confidence interval 0.76-1.37); for women, 1.23 (95% confidence interval 0.96-1.58); group 3 for men, 1.50 (95% confidence interval 1.09-2.06); for women, 1.33 (95% confidence interval 1.03-1.73); and group 4 for men, 2.10 (95% confidence interval 1.53-2.88); for women, 2.25 (95% confidence interval 1.78-2.84). The following standardized mortality ratios were calculated by using the respective geographic area (the Federal State of North Rhine Westphalia) as reference population: group 1 for men, 1.26 (95% confidence interval 0.98-1.61); for women, 1.00 (95% confidence interval 0.81-1.23); group 2 for men, 1.31 (95% confidence interval 1.09-1.57); for women, 1.20 (95% confidence interval 1.02-1.40); group 3 for men, 1.92 (95% confidence interval 1.53-2.38); for women, 1.27 (95% confidence interval 1.07-1.50); and group 4 for men, 3.05 (95% confidence interval 2.47-3.73); for women, 2.31 (95% confidence interval 2.04-2.60). In addition to age, sex, and BMI, Cox proportional hazards models revealed systolic blood pressure, glucose intolerance, diabetes, and smoking as significant independent mortality risk factors, whereas cholesterol was not significant. In this prospective study of a large cohort of obese persons, morbid obesity (BMI of > or = 40 kg/m2) was a strong predictor of premature death. Excess mortality risks associated with gross obesity (BMI from 32 to < 40 kg/m2) were considerably lower than hitherto assumed; moderate degrees of obesity (BMI from 25 to < 32 kg/m2) were not significantly associated with excess mortality.