Ganglioside treatment was evaluated with a multicenter, randomized, double-blind, controlled, cross-over vs placebo trial in 140 insulin-treated diabetic subjects with peripheral neuropathy. The patients entered the study when they showed an impairment in at least two of the electroneurographic parameters, and were assigned to two protocols according to the presence and severity of their neurological symptoms. Ninety-seven diabetic subjects with no or mild symptoms were assigned to protocol I, whereas 43 symptomatic patients were assigned to protocol II. the treatment periods lasted 6 weeks with an intermediate washout period of 4 weeks. The treatment consisted in the daily i.m. administration of 20 mg gangliosides or of placebo. Electroneurographic parameters were recorded at the beginning and at the end of each treatment period, whereas clinical and metabolic data (mean daily plasma glucose, glycosuria and glycosylated hemoglobin) were evaluated every three weeks in protocol I and every two weeks in protocol II. No change in the metabolic parameters was observed throughout the trial period. However, the treatment induced a statistically significant improvement of paresthesias (protocol II) and of some electrophysiological parameters; in particular, ganglioside treatment improved MCV of peroneal nerve (p less than 0.03) in patients of protocol I, MCV o ulnar nerve (p less than 0.002) and SCV of median nerve (p less than 0.06) in patients of protocol II. Furthermore, 22 subjects of protocol II showed a 'drug preference' while 10 preferred placebo and 9 had no preference. In conclusion, ganglioside treatment seems to have a positive effect on diabetic peripheral neuropathy, improving both some symptoms and some electrophysiological parameters.