Much of bladder cancer in East Africa and the Middle East is attributed to chronic urinary infection with Schistosoma haematobium ('schistosomiasis'). Most schistosomal bladder cancer (SBC) is squamous cell carcinoma (SCC) and occurs in the fifth decade of life. In contrast, nonschistosomal bladder cancer (NSBC) in Western countries usually occurs in the seventh decade of life and is largely transitional cell carcinoma (TCC). To shed light on the mechanisms underlying these different patterns of bladder cancer we looked for mutations in the p53 gene in SBC from 92 patients in Egypt, where schistosomiasis is hyperendemic. Patients' mean age at presentation of bladder cancer was 49.4 +/- 9.9 years and 90% had a clinical history of schistosomiasis and/or histological evidence of schistosomal eggs adjacent to the carcinoma. There were 53 SCC, 23 TCC, 13 adenocarcinomas and three other carcinomas. Thirty patients had tumours with mutations in exons 5-8 of the p53 gene: 17/53 SCC, 8/23 TCC, 4/13 adenocarcinomas and 1/3 other tumours. Of 19 mutations in SCC, 16 were base pair substitutions (BPS), two were deletions and one an insertion. Two tumours each contained two mutations. Of the BPS, nine were transitions at CpG dinucleotides and two were G-->T transversions. All the mutations in TCC were BPS: four were transitions at CpG dinucleotides and three were G-->C transversions. One TCC had two mutations. Of four adenocarcinomas with mutations, two had transitions at CpG dinucleotides. Of the 30 BPS mutations, 16 were transitions at CpG dinucleotides, of which 12 were C-->T. We combined these 33 mutations with six obtained from Egyptian SCC reported by Habuchi et al. (Cancer Res., 53, 3795-3799, 1993) to compile a mutational spectrum. This was compared with a NSBC spectrum assembled from 118 mutations reported in the literature. The proportion of BPS at CpG dinucleotides was significantly higher in SBC than in NSBC (18/34 versus 25/103, P = 0.003). There was also a bias away from mutations in exons 7 and 8 towards mutations in exons 5 and 6. We suggest that the excess of transitions at CpG dinucleotides in SBC results from nitric oxide (NO) produced by the inflammatory response provoked by schistosomal eggs. NO could produce such mutations directly, by deamination of 5-methylcytosine, and indirectly, following conversion to nitrate, bacterial reduction to nitrite and endogenous formation of urinary N-nitroso compounds. These produce O6-alkylguanines in DNA, leading to very high rates of G:C-->A:T transitions, a process possibly augmented by inefficient repair of alkylated bases at CpG dinucleotides.