PATIENTS AND METHODS: This prospective, randomized, double-blind study included 65 patients who required surgical removal of impacted mandibular third molars with Class II or position B impaction (Pell and Gregory classification). Patients were randomly assigned to 1 of 3 groups: dexamethasone, methylprednisolone, or placebo (control). Surgery was performed with patients under local anesthesia. Baseline measurements were obtained preoperatively, and subsequent assessments were made on postoperative day 1, 2, 5, and 7 to measure postoperative facial swelling by use of 2 linear measurements: interincisal mouth opening width and visual analog scale score for pain. The amount of analgesics consumed was recorded. Wound healing also was assessed on postoperative day 7. Descriptive and multivariate statistics were computed, and significance was set at P