Variable onset latency of single-injection sciatic nerve block (SNB) may result from drug deposition insufficiently close to all components of the nerve. We hypothesized that this variability is caused by the needle tip position relative to neural components, which is objectified by the type of evoked motor response (EMR) elicited before local anesthetic injection. One-hundred ASA I-II patients undergoing reconstructive ankle surgery received infragluteal-parabiceps SNB using 0.4 mL/kg (maximum 35 mL) of levobupivacaine 0.625%. The end-point for injection was the first elicited EMR: inversion (I), plantar flexion (PF), dorsiflexion (DF), or eversion (E) at 0.2-0.4 mA. The frequencies of the EMRs were: I 40%, PF 43%, E 14%, and DF 3%. SNB was considered complete if both tibial and common peroneal nerves were blocked and failed if either analgesia to pinprick was not observed at 30 min or anesthesia at 60 min. Patients with an EMR of I demonstrated shorter mean times (+/-95% confidence interval [CI]) to complete the block with 8.5 (95% CI, 6.2-10.8) min compared to 27.0 (95% CI, 20.6-33.4) min after PF (P < 0.001) and 30.4 (95% CI, 24.9-35.8) min after E (P < 0.001). No rescue blocks were required in group I compared with 24% (P = 0.001) and 71% (P < 0.001) of patients in groups PF and E, respectively. We conclude that EMR type during nerve stimulator-assisted single-injection SNB predicts latency and success of complete SNB because the observed EMR is related to the positioning of the needle tip relative to the tibial and common peroneal nerves.