Reports concerning duration of action of atracurium in obese patients are conflicting. The aim of this study was to evaluate different anthropometric variables as predictors for duration of action of atracurium-induced block. We studied 127 female patients (total body weight 46-119 kg) anesthetized with midazolam, fentanyl, thiopental, nitrous oxide, and halothane. Twelve different anthropometric variables were evaluated as predictors for duration of action. Linear, least-square, regression analyses were used. There was a significant correlation between each of the 12 variables and the duration of action. The predictors with the greatest correlation coefficients for duration of action of the atracurium induction dose (0.5 mg/kg) were total body weight divided by surface area (r2 = 0.284, P < 0.0001), body mass index (r2 = 0.265, P < 0.0001), and total body weight (r2 = 0.264, P < 0.0001). The most significant predictors for the duration of action of the first supplemental atracurium dose (0.15 mg/kg) were total body weight divided by surface area (r2 = 0.170, P < 0.0001) and total body weight (r2 = 0.160, P < 0.0001). We propose that the atracurium dose should be reduced with 0.23 mg for each kilogram of total body weight above 70 kg. We conclude that the duration of action of atracurium block is prolonged in obese patients, and that atracurium dose in milligrams per kilogram of total body weight should be reduced in these patients. Total body weight divided by the surface area and total body weight were the best predictors for duration of action of atracurium-induced neuromuscular block.