BACKGROUND Quantified, human hemoglobin (Hb)-specific, immunochemical fecal occult blood test (IFOBT) measurements are now used for colorectal cancer (CRC) screening. The objective was to evaluate sensitivity and specificity for CRC and advanced adenomatous polyps (APs) by the fecal Hb threshold used to determine a positive test and the number of IFOBTs prepared per test, so as to determine the least number of colonoscopies required to detect a neoplasm. METHODS Cumulative data were analyzed from a prospective cross-sectional double-blind study of 1682 consecutive, ambulatory, nonbleeding colonoscopy patients who volunteered for IFOBTs, most of above average risk, from 3 ambulatory-endoscopy centers. Fecal Hb was measured in 3 samples and analyzed by an automated instrument, and the highest result >or=50 ng Hb/mL of buffer was related to findings. RESULTS Colonoscopy identified CRC in 20 patients and advanced APs in 129. Sensitivity for either was best when any of 3 tests had >or=50 ng Hb/mL of buffer; sensitivity was 61.1% (95% confidence interval [CI], 53.2-68.9), and specificity was 87.8% (95% CI, 86.2-89.4). Positive tests identified 100% of CRCs and 55% of advanced APs every 3.1 colonoscopies. Sensitivity of a single test at the commonly used 100-ng Hb/mL threshold was lower at 31.5% (95% CI, 24.1-39.0) (P<.001), but specificity was higher at 96.4% (95% CI, 95.5-97.3) (P<.001). Positive tests identified 65% of CRCs and 26.4% of advanced APs every 2.2 colonoscopies. CONCLUSIONS The fecal Hb cutoff chosen by the screener and the number of samples collected per patient determine sensitivity and specificity for CRC/advanced AP; these factors determine the number of colonoscopies needed for positive tests and neoplasia yield. This information provides guidelines for IFOBT screening. Limitations are 1-time screening and most examinees not being at average risk for CRC.