Objective: To examine the relation between different types of alcoholic drinks and upper digestive tract cancers (oropharyngeal and oesophageal). Design: Population based study with baseline assessment of intake of beer, wine, and spirits, smoking habits, educational level, and 2-19 years' follow up on risk of upper digestive tract cancer. Setting: Denmark. Subjects: 15 117 men and 13 063 women aged 20 to 98 years. Main outcome measure: Number and time of identification of incident upper digestive tract cancer during follow up. Results: During a mean follow up of 13.5 years, 156 subjects developed upper digestive tract cancer. Compared with non-drinkers (drinkers of < 1 drink/week), subjects who drank 7-21 beers or spirits a week but no wine were at a risk of 3.0 (95% confidence interval 1.5 to 6.1), whereas those who had the same total alcohol intake but with wine as >30% of their intake had a risk of 0.5 (0.2 to 1.4). Drinkers of > 21 beers and spirits but no wine had a relative risk of 5.2 (2.7 to 10.2) compared with non-drinkers, whereas those who drank the same amount, but included wine in their alcohol intake, had a relative risk of 1.7 (0.6 to 4.4). Conclusion: A moderate intake of wine probably does not increase the risk of upper digestive tract cancer, whereas a moderate intake of beer or spirits increases the risk considerably.