In the absence of any biologic rationale or clinical data for establishing a steroid dosage for suppressing the initial inflammatory manifestations of SLE or for tapering the steroid dosage, this exercise provides a model and empirical data with which to standardize one of the most difficult aspects of SLE clinical trials. The committee views this as a work-in-progress. Studies of steroid prescribing in SLE suggest that physician characteristics are more important determinants than patient characteristics (5); our exercise confirms the variation of dosing. Not standardizing the use of steroids in clinical trials threatens both the internal and external validity of such trials (6). For example, the assembly of a homogenous population needs to take into account baseline steroid use. Since steroids affect every relevant outcome in SLE, including mortality, disease activity, pain, and quality of life, equal distribution of steroid use and steroid dosage is crucial. Because most SLE trials are small, stratified randomization should be considered, since the lack of a statistically significant difference may be the result of small sample sizes and a lack of power. Even with comparable study groups at baseline, differential steroid use postrandomization would likely bias the results, and again, statistical adjustment may not be successful. With small numbers of subjects reporting differential steroid use, quantitative conclusions may not be possible. Detailed predefined protocolized steroid use can eliminate this source of bias but may also lead to dropouts or to noncompliance. Using the response criteria developed by the ACR for global or overall disease activity (4), patients who drop out because of the need for increased steroid dosages can be classified as nonresponders. Future work in this area might profitably examine the choice of the initial steroid and tapering regimen on outcomes, the use of this scheme in the separation of therapies in clinical trials, and the determinants of both the dosage and duration of steroids in specific SLE phenotypes.