Epidemiological studies on the relation between coffee consumption and cancer risk have been mainly focused on cancers of the urinary bladder, pancreas and colorectum. The relation between coffee and bladder cancer is controversial, despite a large number of studies published over the last three decades. In most studies, the risk tends to be higher in coffee drinkers than in those who do not drink coffee, but the excess risk is generally moderate and is neither dose- nor duration-related. Thus, a strong association between coffee drinking and bladder cancer can be excluded, although it is still unclear whether the weak association is causal or nonspecific and due to some bias or confounding. For pancreatic cancer, a possible association with coffee consumption has been postulated in a large case-control study published in 1981; since then, however, most studies have shown no substantial association, and overall evidence suggests that coffee is not materially related to pancreatic cancer risk. Overall evidence on the coffee-colorectal cancer relation suggests an inverse association, since most case-control studies found odds ratios below unity, particularly for colon cancer. The pattern of risk is less clear for cohort studies. A plausible biological explanation has been given in terms of coffee-related reduction of bile acids and neutral sterol secretion in the colon. For other cancer sites, including oral cavity, oesophagus, stomach, liver, breast, ovary, kidney and lymphoid neoplasms, the relation of coffee drinking with cancer risk has been less extensively investigated, but the evidence is largely reassuring.