Three techniques for blocking the sciatic nerve, differing in approach (posterior in group P; lateral in group L; and anterior in group A), were prospectively evaluated in 180 children who were also given light general anesthesia for surgery below the knee. Four anesthetic solutions with epinephrine (1% lidocaine, 0.5% bupivacaine, and two mixtures of 0.5% bupivacaine with either 1% lidocaine or 1% etidocaine) were administered to 15 patients in each group. The sciatic nerve was located by electrical stimulation or, when muscle twitches were not elicited, using a loss-of-resistance technique. Twitches were "typical" in 154 patients of whom 153 developed sciatic nerve block. In 26 patients twitches were atypical (eight patients) or absent (18 patients) and a sciatic block developed in only 13 patients (50%). The depth to which the needle was inserted was measured in each procedure; it varied according to patient's age and weight and was significantly less with the posterior approach than with either the lateral or anterior routes. The overall success rate exceeded 90% in the three groups but significantly fewer difficulties were encountered in group P than in group A. Although the spread of the anesthetic was different in the three groups, the distribution of anesthesia in the lower extremity was similar, including not only dermatomes supplied by the sciatic nerve, but also those supplied by the posterior femoral cutaneous nerve. No neurological sequelae were observed. It is concluded that the posterior and lateral approaches are the most suitable in children for blocking the sciatic nerve proximally.