Abstract.Objective. Based on a meta‐analysis, it was recently stated that there is no association between coffee consumption and the risk of coronary heart disease. Why then, have studies on the issue shown quite variable results?Design, setting and subjects. A prospective study was performed in the Copenhagen Male Study on 2975 men (53–74 years) without cardiovascular disease at the baseline in 1985/1986. They were classified according to self‐reported consumption of filter coffee. Some 147 men (5%) were coffee abstainers. Potential confounders were alcohol use, physical activity, smoking, serum cotinine, serum lipids, serum selenium, body mass index, blood pressure, Lewis blood group, hypertension, non‐insulin‐dependent diabetes mellitus and social class.Main outcome measures. The incidence of ischaemic heart disease (IHD) 1985/86–1991.Results. Some 184 men had a first IHD event. There was no significant difference between those consuming 1–4, 5–8 or ≥ 9 cups per day after controlling for confounders (P‐value of trend test: 0.14). The crude incidence rates were 6.8, 6.7 and 4.6%, respectively; the adjusted rates were 6.8, 6.7 and 4.0%, respectively. Coffee consumption was significantly (P<0.05) inversely correlated with serum selenium concentration (never previously described) and, positively or negatively, with a number of other potential risk factors: smoking, alcohol use, serum triglycerides, serum cholesterol, blood pressure, social class, body mass index, and serum selenium. In nonsmokers and smokers of only a small amount of tobacco, coffee consumption was associated with a lower risk of IHD (P<0.05).Conclusion. We conclude that the association between coffee consumption and risk of IHD is conditioned by known risk factors correlated with use of coffee, which may partly explain the inconsistencies in the results