In a population study for colorectal neoplasia possible relationships were sought between faecal occult bleeding and clinical findings. Samples were taken from three different stools, collected during biennial screening from 1985 to 1994 in the age group 45–75 years at the beginning of the study. A total of 943 persons had positive Haemoccult-II (H-II) tests in at least one of the five screening rounds, and 87% had a complete colonoscopy. Of the 20,672 persons accepting initial screening, 1% had a positive test, the figure increasing to 1.8% during the fifth screen. The degree of positivity (1–3 slides) decreased with time. Positive tests were most frequent in men. Predictive values (PVpos) for colorectal cancer (CRC) increased with increasing number of positive slides from 6 to 31%. The average PVposdecreased from 17% during the initial screen to 8% during the fifth. Average PVposfor possible precursors (adenomas) varied from 40% at the beginning to 33% at the end. PVposfor CRC and inflammatory bowel disease increased when visible blood in the stools was present. Marcoumar increased PVposfor adenomas. Colorectal examination in the 943 persons revealed 114 persons with CRC and further 377 with adenomas. Early CRC (Dukes' stage A) was detected in 46 of the 114 with CRC, a more favourable distribution than that found in symptomatic patients. The study defined a group of people with a high risk of colorectal neoplasia, with a good prognosis, out of a group who had been selected at random from a normal population.