Summary Two faecal occult blood tests (FOBTs), Hemoccult II (guaiac based) and Hemeselect (immunochemical) were compared in a population screening for colorectal cancer on 24 282 subjects aged 40-70. Hemeselect was interpreted according to a lower (+ and +) and a higher (+) positivity threshold. A total of 8008 compliers were enrolled in the study. Positivity rates: Hemoccult=6.0%, Hemeselect (+ and ±) = 8.2%, Hemeselect (+) = 3. 1%. Among FOBT-positive subjects complying with the diagnostic work-up, 22 had colorectal cancer (17 Hemeselect-positive (+), four Hemeselect-borderline (±), 15 Hemoccult-positive) and 166 subjects had adenomas (62 Hemeselect(+), 56 Hemeselect-borderline (±), 79 Hemoccult-positive) were detected. The positive predictive values (PPVs) for cancer were as follows: Hemoccult =3.7%, Hemeselect (+ and +) = 3.8%, Hemeselect (+) =8.4%. The PPVs for adenoma(s) were: Hemoccult = 19.7%, Hemeselect (+ and +) =21.4%, Hemeselect (+) =30.5%. The specificity for cancer was: Hemoccult =94.1%, Hemeselect (+ and +) = 92%, Hemeselect (+) = 97.1%. Ratios between detection rates of each test and expected incidence of colorectal cancer suggest that Hemoccult anticipates cancer diagnosis by approximately 2 years on average whereas the mean diagnostic anticipation of Hemeselect ranges between 2.5 and 3.2 years. Hemeselect is superior to Hemoccult as it is at least as effective but more efficient and acceptable than guaiac testing. Further evaluation of Hemeselect cost-effectiveness and sensitivity is needed in order to assess the optimal threshold of positivity and screening frequency.