Based on the analysis of 13,110 medical examinations performed on a standardized population of pneumoconiosis patients recorded on the F.O.D. computer file, the authors describe the value of the subjective estimations of »obesity«, »thinness« or »normal weight« by their correlation with the observed weight and height. Although there are striking differences in appreciation between the physicians performing the examinations, the qualifications »obese«, »thin« or »normal« correspond with real group differences in weight, between certain limits which can be defined. The ratio between the observed weight and the expected weight (using the Broca formula) shows the same pattern. In tins way it becomes possible to propose upper and lower limits for obesity, thinness and normal weight based on purely empiric data. Feeding back this information to the examining physicians should help reduce the differences between physicians and improve the results. Therefore, the authors find it useful to keep such information in the computer file.