A retrospective study was carried out, over a twelve-year period, of all cases of acute chloroquine poisoning where more than 2 g of chloroquine had been taken. It included 386 patients; of these, 60 who had taken drugs other than chloroquine, and 17 who had ingested less than 1 g of the drug, were excluded. The remaining 309 patients were allocated to two groups: a "control group", consisting of the patients admitted between January 1973 and April 1980 (n = 146), and a "diazepam group", made up of those admitted from May 1980 to December 1989 (n = 163). The patients in the latter group had had the same symptomatic treatment as those in the control group, and had been routinely given a 0.5 mg.kg-1 bolus of diazepam on admission followed by 0.1 mg.kg-1.day-1 for every 100 mg of chloroquine supposed to have been ingested. Both groups were divided into three subgroups, those patients with cardiorespiratory arrest, and those with, and those without, symptoms on admission. No statistically significant difference was found between either the control and diazepam groups or between subgroups, concerning the distribution of age, sex, amount of chloroquine supposed to have been ingested, delay in hospital admission and death rate. However, there was a higher death rate in the asymptomatic subgroup not treated with diazepam than in the diazepam group. Therefore, the routine use of diazepam for the treatment of acute chloroquine poisoning does not seem to be justified in symptomatic cases and in those with inaugural cardiac arrest.(ABSTRACT TRUNCATED AT 250 WORDS)