High cyanide intake from consumption of insufficiently processed cassava has been advanced as a possible aetiology of the upper motor neurone disease konzo. However, similar neurodamage has not been associated with cyanide exposure from any other source. With an ecological study design, we compared 22 cases of konzo, 57 unaffected household members and 116 members from unaffected households, a total of 195 subjects, in konzo-affected savanna villages with 103 subjects in adjacent non-affected forest villages in the Paykongila area in the Bandundu Region, Zaire. In the dry season, the mean value (+/- SEM) of urinary thiocyanate, the main cyanide metabolite, was higher in the three groups in konzo-affected villages (563 +/- 105, 587 +/- 44 and 629 +/- 47 micromol/l) than in unaffected villages (241 +/- 17 micromol/l). In affected villages in the dry season when konzo incidence was high, mean urinary thiocyanate was also higher than the levels found in the wet season when incidence was low. The wet season values (mean +/- SEM) were 344 +/- 60, 381 +/- 35 and 351 +/- 27 micromol/l. Urinary levels of inorganic sulphate were low in all groups, indicating low intake of the sulphur amino-acids which provide a substrate for cyanide detoxification. These findings support an aetiological role for cyanide in konzo. However, urinary linamarin, the cyanogenic glucoside and source of cyanide in cassava, was more closely associated with the occurrence of konzo. The mean value (+/- SEM) of urinary linamarin in the konzo cases was 632 +/- 105 micromol/l and in their household members 657 +/- 52 micromol/l, which was significantly higher than in members of control households in the same village (351 +/- 28 micromol/l) and in unaffected villages (147 +/- 18 micromol/l). This suggests that a specific neurotoxic effect of linamarin, rather than the associated general cyanide exposure resulting from glucoside breakdown in the gut, may be the cause of konzo.