To estimate the efficacy of screening on colorectal cancer mortality, a population-based case-control study was conducted in well-defined areas of Burgundy (France). Screening by faecal occult blood test prior to diagnosis in cases born between 1914 and 1943 and who died of colorectal cancer diagnosed in 1988-94 was compared with screening in controls matched with the case for age, sex and place of residence. Cases were less likely to have been screened than controls, with an odds ratio (OR) of 0.67 [95% confidence interval (CI) 0.48-0.94]. The negative overall association did not differ by gender or by anatomical location. The odds ratio of death from colorectal cancer was 0.64 (95% CI 0.46-0.91) for those screened within 3 years of case diagnosis compared with those not screened. It was 1.14 (95% CI 0.50-2.63) for those screened more than 3 years before case diagnosis. There was a negative association between the risk of death from colorectal cancer and the number of participations in the screening campaigns. The inverse association between screening for faecal occult blood and fatal colorectal cancer suggests that screening can reduce colorectal cancer mortality. This report further supports recommendations for population-based mass screening with faecal occult blood test.