We used 20 ml ropivacaine 0.75% for ankle blocks before foot surgery in 90 participants who we allocated in equal numbers to: perineural dexamethasone 8 mg and intravenous saline 0.9%; perineural saline 0.9% and intravenous dexamethasone 8 mg; or perineural and intravenous saline 0.9%. Dexamethasone increased the median (IQR [range]) time for the return of some sensation or movement, from 14.6 (10.8–18.8 [5.5–38.0]) h with saline to 24.1 (19.3–29.3 [5.0–44.0]) h when given perineurally, p = 0.00098, and to 20.9 (18.3–27.8 [8.8–31.3]) h when given intravenously, p = 0.0067. Dexamethasone increased the median (IQR [range]) time for the return of normal neurology, from 17.6 (14.0–21.0 [9.5–40.5]) h with saline to 27.5 (22.0–36.3 [7.0–53.0]) h when given perineurally, p = 0.00016, and to 24.0 (20.5–32.3 [13.0–42.5]) h when given intravenously, p = 0.0022. Dexamethasone did not affect the rates of block success, postoperative pain scores, analgesic use, or nausea and vomiting. The route of dexamethasone administration did not alter its effects.