A case-control study of oral cancer was conducted in Beijing, People's Republic of China (PRC). The study was hospital-based and controls were hospital in-patients matched for age and gender with the cases. The response rates for cases and controls were 100 percent and 404 case/control pairs were interviewed. Tobacco smoking and alcohol consumption emerged as independent risk factors for oral cancer. For tobacco smoking, the association was considerably stronger for smokers of pipes than for smokers of cigarettes. For all kinds of tobacco, expressed as cigarette equivalents, the odds ratio (OR) for total pack-years smoked, among males, rose from 1.0 in never-smokers to 3.7 (95 percent confidence interval, 1.8-7.4) in the highest quintile of exposure. Similar results were found for females. The association with tobacco consumption was strong for squamous cell carcinoma but there was no trend in risk associated with tobacco for adenocarcinomas and other histologic types. So few women reported consuming alcohol that this variable could be examined only in males. Risk in the highest category of total lifetime intake of alcohol relative to that in lifetime abstainers was 2.3 (1.1-4.8) with a significant trend in risk with increasing dose (P less than 0.002). The combined effects of tobacco and alcohol appear to be approximately multiplicative in males. The attributable risk of oral cancer for tobacco among tobacco smokers was estimated as 34 percent (45 percent among males and 21 percent among females); for alcohol consumption in males the estimate was 23 percent.