The possibility of screening the general population for prostate cancer using serum prostate-specific antigen (PSA) level (alone or in combination with other tests) as screening test has recently been discussed. A number of studies are on the way, but the published reports have almost exclusively been based on men volunteering for screening. We assessed the feasibility of a screening study based on men identified from a central population registry. A random sample of 600 men in the age groups 55, 60 and 65 years was identified from the Finnish Population Registry as the study population. Half of them were randomised to the intervention group and an invitation to participate was sent to them. The participation rate was 77% (230 out of 300). Twenty-five men had a serum PSA concentration of 4.0 micrograms l-1 or above and were invited for further examination including digital rectal examination, transrectal ultrasound and transrectal Tru-cut biopsies (directed and/or random). Six cases of cancer were detected among the 230 participating men, which corresponds to a detection rate of 2.6% and a positive predictive value of 24%. The number of cases detected is equivalent to the expected number of prostate cancer cases during a 10 year follow-up in this population. The ratio of free to total PSA was also measured and a cut-off level of 0.20 was chosen. Its use as an additional criterion of the screening test would have decreased the prevalence of false-positive screening tests from 8% (19 of 230) to 3% (7 of 230) at a cost of missing one of the six cancers compared with serum total PSA concentration alone. Of the six cancers, five were clinically regarded as localised and locally confined disease was confirmed pathologically in four of them. In conclusion, a population-based study in Finland seems feasible and the properties of the PSA test can be regarded as suitable for a randomised screening study. Thus, all prerequisites for a multicentre study, which is planned, seem to exist.