To appraise the role of physical activity in reducing coronary mortality among longshoremen, 6351 men, 35 to 74 years old upon entry, were followed for 22 years or to death or to the age of 75. Their longshoring experience was computed in terms of work-years according to categories of high, medium and low caloric output. Individual work assignments were reclassified annually to allow for effect of job transfers. The age-adjusted coronary death rate for the high-activity category was 26.9 per 10,000 work-years, and the medium and low catgories had rates of 46.3 and 49.0 which were little different from each other. This protective "threshold" effect was seen especially for the sudden-death syndrome, in which the death rate for heavy workers was 5.6, as contrasted with 19.9 for moderate and 15.7 for light workers. We conclude that repeated bursts of high energy output established a plateau of protection against coronary mortality, and that several different mechanisms may explain this finding.