Symptomatic diabetic sensorimotor polyneuropathy is considered progressive and irreversible. The hypothesis that symptoms of diabetic neuropathy may be due to entrapment of peripheral nerves was investigated in a prospective study from 1982 to 1988 in which diabetics (38 type I, 22 type II) had surgical decompression of 154 peripheral nerves in 51 upper extremities and 31 lower extremities. Mean postoperative follow-up was 30 months (range 6 to 83 months). Considering the entire series, an excellent final result was noted for motor function in 44 percent and for sensory function in 67 percent of the decompressed nerves. Ten percent of the patients were not improved, and 2 percent were worse in sensorimotor function. Upper extremity nerve decompressions achieved better results than lower extremity nerve decompressions. Improvement in postoperative electrodiagnostic studies varied in relationship to the preoperative electrodiagnosis. Improvement was noted in 100 percent of those nerves with the preoperative diagnosis of "localized entrapment," 80 percent for "peripheral neuropathy with superimposed entrapment," and 50 percent for "peripheral neuropathy." Progressive neuropathy occurred in a nontreated limb of 50 percent of those patients whose surgically treated limb maintained improvement. The results of this study suggest that symptoms of sensorimotor diabetic neuropathy may be due partly to compression of multiple peripheral nerves. The results further suggest that surgical decompression of such nerves may result in symptomatic improvement.