In an out-patient weight loss study of 63 patients (54 female, 9 male), 53 completed a 16 week treatment with a low calorie diet and a 9 g/day fibre supplement. In these 53 patients, the average weight loss was 8.3 kg (s.e.m. 0.8). Waist-hip ratio (WHR) and abdominal sagittal diameter (SagD) were measured as indicators of fat distribution and visceral adipose tissue (visceral AT) was estimated by anthropometric computerized tomography calibrated equations. Four observers measured WHR and SagD ten times in eight patients. Two dietitians examined the patients throughout the clinical trial at weeks 0, 4, 8 and 16. Furthermore, two physicians examined the patients at week 12 in the trial. Two- and three-way analyses of variance were performed to estimate the contribution of single factors to the total variance. The contribution of observers, 3.2% and 3.8%, respectively, was of the same magnitude as the error variance (2.9% and 4.8% respectively) which is a measure of the intra-observer variation. The two dietitians had very similar recordings and contributed only 0.3% and 0.9% to the total variance for WHR and SagD, respectively. The contributions of the two physicians to the total variance were 0.0% for WHR and 0.4% for SagD. It is concluded that there is no need to use several observers or repeated measurements of waist, hip and SagD in clinical anti-obesity trials.