The main purpose of population screening for colorectal cancer is to reduce mortality from the disease. The criteria of death from colorectal cancer are defined in the present randomized trial of 61,938 persons between 45 and 74 years old, and the need for an impartial death review committee was demonstrated. Causes of death within the first 5 years are described within subgroups of the test group and compared with those in the control group. Death rates were higher among non-responders to screening than among controls and among those in whom Hemoccult-II had been performed at least once. Persons with negative Hemoccult-II had a lower death rate than controls. The overall autopsy rate was 32%. Lethal complications from treatment of colorectal neoplasia were evaluated per se. Death from colorectal cancer occurred in 74 persons in the total screening group and in 91 among controls. Sources of bias are discussed. A method of evaluating possible benefit to those being screened is suggested. Final results cannot be expected before 1996.