This study examines three faecal occult blood tests, Haemoccult, Fecatwin and E-Z Detect, each with different sensitivities, to determine which is best suited for use in symptomatic patients--both for the detection of cancer and of non-malignant mucosal disease of the large bowel. A test was completed by 1025 patients before double-contrast barium enema and the performance of each test was determined from the result of this investigation. The study was completed by 969 patients. There were 49 patients with colorectal cancer, 92 patients with a cancer or a polyp greater than 5 mm, and 130 with some mucosal abnormality. The test most sensitive for blood, Fecatwin, detected 14 of 15 (93 per cent) cancers and 29 (69 per cent) of 42 patients with mucosal disease (including inflammatory bowel disease) but gave three times as many false positive results as the Haemoccult test, which is less sensitive for blood. The chance of a patient with a positive Haemoccult result having mucosal disease on barium enema was 24 of 47 patients (51 per cent) (two-thirds of these having colorectal cancer). A negative Haemoccult result, however, was unreliable and should not influence patient management. A test less sensitive for blood than Haemoccult was found to be of little value in symptomatic patients.