A study was conducted in northern Iran in areas of very high, high and moderately low incidence of oesophageal cancer. Morphine metabolites in urine as an indicator of opium use, and a variety of nutritional and biochemical measures, including salivary antipyrine half-life, were determined in households with a case of oesophageal cancer and in control households. Results on 1,590 individuals showed that the prevalence of appreciable levels (greater than or equal to 1 microgram/ml) of urinary morphine metabolites was much higher in areas of high and very high incidence of oesophageal cancer than in low-incidence areas, particularly for those under age 50 years in both sexes, where a 6-fold difference was seen. Members of households with a case of oesophageal cancer had a higher prevalence of positive urinary morphine metabolite findings than members of control households from the same village. Salivary anti-pyrine half-life (AP-T1/2), as a measure of hepatic drug metabolizing capacity, was estimated in 120 subjects. No relationship was seen with use of opium, region of residence, case-control status of household, age or sex. Experimental studies have meanwhile identified mutagenic substances with carcinogenic potential in opium pipe scrapings, used widely in the study region, opium pyrolysates and morphine pyrolysates. Our data give additional support to the hypothesis that opium use, in the form of its pyrolysates, is one of the factors involved in the aetiology of oesophageal cancer in the region.