The author reviews the literature on occult blood surveillance for colorectal carcinoma. The guaiac-based Hemoccult (SmithKline Diagnostics, Sunnyvale, Calif.) test is the most reliable and widely used. However, testing is complicated by several technical issues that can affect clinical results, and even successful screening programs will miss a high proportion of tumors. Public compliance is often poor, and a number of indirect and "hidden" costs make surveillance programs much more expensive than is usually claimed. Almost all published screening trials are uncontrolled. They generally detect about 3-20 colorectal malignancies for every 10,000 people enrolled, but only about 5%-10% of occult blood reactions are due to cancer. Though screen-detected tumors tend to be at a relatively early stage, this does not imply any benefit of surveillance because of lead time and length biases inherent in the screening process. Only controlled trials can answer the central question of whether screening decreases mortality from bowel cancer. Two such trials are underway, but mortality data are not yet available from either.