All the men living in Malmö born in 1926-9 were invited for a screening examination which included an assessment of alcohol consumption and measurement of gamma-glutamyltransferase (GGT) activity. They were followed for up to four years (median 2) and their mortality assessed. Sixty-two deaths occurred, 41 (0.9%) among the 4571 men who attended the screening investigation and 21 (1.3%) among the 1609 who did not respond to the invitation. Evidence of alcohol abuse or an alcohol-related cause of death was present in 25 (61%) of the deaths among the attenders and 13 (62%) of those among the non-responders. GGT values at the screening investigation were significantly increased in 19 (46%) of those who died, but established risk factors, such as cholesterol and triglyceride concentrations and blood pressure, had little predictive value. Measurement of GGT provided an objective index of alcohol consumption, though the full clinical importance of a raised value needs further assessment. The finding that heavy alcohol consumption was the single most important factor associated with premature death in these middle-aged men has important implications for prevention.