A retrospective case-control study of 713 male oral-cancer patients seen at Tata Memorial Hospital, Bombay, during 1980-1984 was undertaken to assess the association between chewing, smoking and alcohol habits. Male controls were chosen among those persons who attended the hospital during the same period and were diagnosed as free from cancer, benign tumour and infectious disease. Statistical analysis was based on unconditional logistic regression and the confidence interval for RR was calculated using the standard error of the estimates. Established factors such as tobacco chewing and bidi smoking showed a significant association with oral cancer. For the alcohol habit, the relative risk was 1.42 and the dose-response relationship, in terms of frequency and duration of the habit, was also observed. The illiterate group showed an almost 2-fold significant excess risk compared to the literate group. After adjusting for confounding variables such as age, residence, illiteracy and known factors such as tobacco chewing and bidi smoking, the study has brought out the significance of a non-vegetarian diet as a high-risk factor for oral cancer compared to a vegetarian diet. Further studies are required to identify specific items in the non-vegetarian diet which may be associated with oral cancer.