Experience using three local anesthetic agents (lidocaine hydrochloride, chloroprocaine hydrochloride, and prilocaine) intravenously in producing regional anesthesia in extremities isolated from the systemic circulation by a tourniquet is compared. Chloroprocaine produced thrombophlebitis in 4 of 51 patients. Although prilocaine produced excellent anesthesia, its use should be reserved for clinical trials until further data are available concerning the production of methemoglobinemia. Taking both effectiveness and safety into account, lidocaine appears to be the drug of choice at this time. In our experience, and from the reports in the literature, 1.5 mg/kg of lidocaine hydrochloride is preferable to 3 mg/kg. This dose resulted in good or excellent regional anesthesia in 60 of 66 patients when injected after a 20-minute period of ischemia had been effected by the use of an arterial tourniquet.