Background: We have previously reported how the SCORAD index was designed. This cumulative index combines objective (extent and intensity of lesions) and subjective (daytime pruritus and sleep loss) criteria. Aims: To study interobserver variability in scoring for objective SCORAD criteria and to optimize the scoring guidelines. Material and Methods: Three scoring sessions were organized in 1993-1994 in Hamburg, Bordeaux and Rotterdam totalizing 19 patients (14 children and 5 adults) and 23 physicians, among whom 12 participated in at least 2 scoring sessions; 169 evaluation sheets have been processed using the SCORAD File Marker Pro software. At each session, total body photographs and close-up views were taken of each patient, and this material was reviewed at the final evaluation. Results: The extent of lesions according to the rule of nines showed interobserver variability mostly for patients with lesions of moderate intensity involving 20–60% of body surface. Intensity items were scored with more consistency overall, but variations subsided especially for oozing/crusts and lichenifications. Low and high scorer profiles and the benefit of training were noted. Conclusions: This study has allowed to optimize clinical scoring using the SCORAD system. A proposal has been made to grade the severity of atopic dermatitis according to objective criteria in three groups for inclusion in clinical trials. The SCORAD index remains the major criterion for follow-up in trials.