A case-control study of oesophageal cancer was carried out in Trivandrum, Kerala, involving 267 cases and 895 controls. Risk factors studied in males were pan (betel)-tobacco chewing, bidi and cigarette smoking, drinking alcohol and taking snuff. Only pan-tobacco chewing was investigated in females as very few indulged in the other habits. Among males significant associations with higher risk were observed for bidi smoking (p less than 0.001), bidi plus cigarette smoking (p greater than 0.05) and drinking alcohol (p less than 0.001). While a significant effect of duration of pan-tobacco chewing (p less than 0.005) was observed in males, there was no significant trend, the risk first falling then rising as duration of use increased. This was partly due to confounding with smoking. No effect of pan-tobacco use was observed in females. A step-wise model was fitted, retaining only those risk factors which were significant when adjusted for other factors; the risk factors included were duration of pan-tobacco chewing, duration of bidi smoking, daily frequency of bidi and cigarette smoking and alcohol use (yes or no). An adjusted relative risk of 2.03 was observed for a pan-tobacco habit of more than 40 years' duration, of 4.70 for more than 20 years of bidi smoking, of 4.80 for more than 20 bidis/cigarettes per day, and of 2.33 for regular alcohol use (in each category relative to a baseline of those never indulging in the relevant habit).