We estimated the sensitivity of serum prostate-specific antigen (PSA) as a screening test for prostate cancer in the Finnish randomised, population-based prostate cancer screening trial. The study population consisted of 80,458 men aged 55–67 years identified from the national population registry and randomised to the screening or control arm of the trial. The screening algorithm was based on determination of serum PSA concentration. Test sensitivity was estimated based on interval cancer incidence during the first 4 years of follow-up among screening participants with a negative screening test. Interval cancers were defined as those occurring among men with a negative screening test. Altogether, 19 interval cancers were detected among 17,897 men with serum PSA < 3 ng/ml during the first screening interval. A further 5 cases were diagnosed among 811 men with PSA 3.0–3.9 ng/ml with a benign digital rectal examination or free total PSA ratio ≥ 0.16. Test sensitivity based on serum PSA of 3 ng/ml was estimated to be 0.89 (95% confidence interval 0.84–0.93) and that based on PSA of 4 ng/ml combined with an ancillary test (digital rectal examination or free total PSA ratio in the PSA range 3.0–3.9) was 0.87 (0.82–0.92). Test sensitivity achieved with serum PSA in prostate cancer screening appears excellent in the context of a population-based effectiveness trial. © 2004 Wiley-Liss, Inc.