Ｏｂｊｅｃｔｉｖｅ　 To explore the effect of non-standardized BMI on subarachnoid anesthesia.　Methods　A total of 100 inpatients with either low or high BMI subjected to lower extremity surgery were enrolled, and were assigned into group A(BMI<20, n=50) and group B(BMI>24, n=50). The patients received subarachnoid injection of 2 ml 0.75% L-bupivacaine through the space between vertebra L3 and L4. The spinal levels of subarachnoid anesthesia were measured by acupuncture and alcohol swab. We analyzed the efficiency of subarachnoid anesthesia in two groups of patients by recording the speed and spinal levels of anesthesia, scores for neuromuscular blocking, HR, BP, time spent for recovery from anesthesia, and adverse reactions to anesthetics.　Results　Although subarachnoid anesthesia in group A developed as fast as that in group B, the vertebra levels of anesthesia in group B(T6, T7) were lower than those in group A(T9, T10). During anesthesia, patients in group A and B did not differ in the scores for neuromuscular blocking, but patients in group A showed less changes in blood pressure and HR. However, patients in group B recovered faster from anesthesia than those in group A, but had higher incidence of adverse reactions to anesthesia.　Conclusions　BMI may affect the spinal levels, recovery, and adverse reactions of subarachnoid anesthesia, and should be taken into consideration in the performance of subarachnoid anesthesia.