Tricyclic antidepressants are the mainstay of treatment of painful polyneuropathy but cannot be used in a substantial number of patients. St. John's wort is a herbal antidepressant, which may act via mechanisms similar to the tricyclics. The aim of this study was to test if St. John's wort would relieve painful polyneuropathy. The study design was randomized, double-blind, placebo-controlled and cross-over. Fifty-four patients were assigned to one of the two treatment sequences. The daily dose of St. John's wort was three tablets each containing 900 microg totalhypericin. During the two treatment periods of 5 weeks duration, patients rated constant pain, lancinating pain paroxysms, touch-evoked pain and pain on pressure by use of 0--10 point numeric rating scales. Forty-seven patients -- 18 diabetics and 29 non-diabetics -- completed the study. There was a trend of lower total pain score (sum of the individual pain scores) on St. John's wort than on placebo (median 14 vs. 15, P=0.05). None of the individual pain ratings were significantly changed by St. John's wort as compared to placebo (P=0.09--0.33). Complete, good or moderate pain relief was experienced by nine patients with St. John's wort and two with placebo (P=0.07). In conclusion, St. John's wort has no significant effect on pain in polyneuropathy.