A single clinical and laboratory assessment from each of 50 randomly chosen patients with systemic lupus erythematosus followed in a prospective study were evaluated for disease activity by 3 individual rheumatologists. Of the 50 assessments, 24 were considered to be active, 12 possibly active and 14 inactive. The 38 assessments that were clearly active or inactive were then analyzed. Clusters of variables were chosen for clinical relevance and association with activity, and 7 highly associated variables were combined into the lupus activity criteria count. Analysis of these criteria in the 50 assessments revealed that the presence of any 2 correctly predicted active disease in 100% of cases. This activity criteria count was then validated using a second sample of 50 assessments.