Objective To compare the effects of the interscalene brachial plexus block performed by three different methods of nerve localization-facial pop（FP）, paresthesia（PAR）, peripheral nerve stimulation（PNS）and the distribution of local anesthetics. Methods 90 patients scheduled for upper extremity operations were randomized to three groups（n=30 in each group): group FP, group PAR and group PNS. The extent of sensory and motor blockade of each innervated region was assessed by scale scores, the anesthetic effects of surgical field and the complications were observed and recorded. Computed tomography（CT）was conducted at transections （C4-T3）and at sagittal and coronal planes of the interscalene injected with anesthetic（6 cases in each group, randomly）. Results There was no significant difference in sensory and motor scale scores of axillary nerve, musculocutaneous nerve, median nerve and musculospiral nerve. The sensory scale scores of cutaneous antebrachii medialis nerve and cubital nerve and the motor scores of cubital nerve were significantly higher in group FP than in group PAR（P<0.05）and in group PNS（P<0.01）, also those in group PAR were significantly higher than in group PNS（P<0.05）. The rate of injury of blood vessel in group PAR was significantly higher than in group FP（P<0.01）and PNS（P<0.05）. The diffusion of local anesthetic in brachiplex sheaths is uneven, and only at interscalene level the sheaths seem to appear，with visible septation and sacs as well as parapodiums. The general anesthetic effect in PNS was significantly better than in group FP（P<0.05），the anesthetic effect of ulnar operations in group FP was significantly worse than in PAR（P<0.05）and in PNS（P<0.01）. Conclusion all of the three methods is advisable in upper arm and radialis operations，while FP is more convenient. In ulnar operations PAR and PNS should be applied, and PNS is more effective.