Tricyclic antidepressants and anticonvulsants have become the mainstay in the treatment of pain in polyneuropathy. Within the last decade, controlled trials have shown that numerous other drugs relieve such pain. To estimate the efficacy of the different treatments, the authors identified all placebo-controlled trials and calculated numbers needed to treat (NNT) to obtain one patient with more than 50% pain relief. The NNT was 2.6 for tricyclic antidepressants, 6.7 for selective serotonin reuptake inhibitors, 2.5 for anticonvulsant sodium channel blockers, 4.1 for the anticonvulsant calcium channel blocker gabapentin, and 3.4 for the mixed opioid and monoaminergic drug tramadol, as calculated from a sufficiently large number of patients. Favorable point estimates of NNT of 1.9 for the NMDA-antagonist dextromethorphan and 3.4 for L-dopa were determined from a limited number of data. For capsaicin, the NNT calculated from many exposed patients was 5.9, but most of the data are controversial owing to trial methodology. Finally, the NNT for the antiarrhythmic sodium channel blocker mexiletine was 38, but this value may be biased because of a lack of dichotomous data in several positive trials. Tricyclic antidepressants are at the moment still the drugs of first choice, and drugs such as gabapentin, carbamazepine, and tramadol may be tried if contraindications or tolerability problems are encountered with the tricyclics.