National Institutes of Health (NIH) Consensus Development Conference, Effect of Corticosteroids for Fetal Maturation on Perinatal Outcomes, was convened to resolve ongoing uncertainties in the medical community on the benefits of administering corticosteroids to women in premature labor. Therapies with uncertain benefits tend to have more varied rates of use across small areas than therapies with established benefits.' A recent paper on rates of use of obstetric and neonatal interventions confirmed that the use of corticosteroids is highly variable: rates of use of antenatal corticosteroids for white low birth weight infants ranged from near 0% to about 58% across 30 centers participating in the Vermont-Oxford Trials Network in 1990.(2) Although random variations in the use of therapies are an indicator of differences in physician practice styles, consistent variations in rates of use provide indicators of (1) areas of agreement in use of the therapy across physicians and (2) practice contexts that either facilitate or generate obstacles to adoption of the therapy. We examined five of the multicenter data sets that were used at the NIH Consensus Conference to assess outcomes of corticosteroid use, along with three additional multicenter data sets, to identify consistencies in the pattern of use of antenatal corticosteroids in premature labor.