OBJECTIVE: To compare the readability and sensitivity of a new guaiac faecal occult blood test, HemoccultSENSA, with those of a standard guaiac-based test, Hemoccult, in a normal working environment. DESIGN: The two tests were performed in parallel on routine clinical and contrived faecal specimens; those developing the tests were blinded as to the test type. SETTING: All tests were carried out in the hospital ward environment under normal conditions by nurses working in the ward. SAMPLES: Fifty faecal samples from healthy volunteer subjects (low concentrations of haemoglobin were added to 40 of these samples) and 145 faecal samples from 65 inpatients likely to have gastrointestinal bleeding. MAIN OUTCOME MEASURES: Test positivity rate, and graded measures of colour intensity, colour stability and colour pattern. RESULTS: With patients' samples, the new test gave a greater number of positive results than the standard test (73.1% v. 65.5%; 95% confidence interval of the difference, 3.3%-11.9%). With contrived samples, the blue colour produced during development was more intense (P less than 0.0003), more stable (P less than 0.0025) and covered a larger area (P less than 0.01) with the new test compared with the standard test. CONCLUSIONS: These results demonstrate the better readability and slightly higher sensitivity of the new test. They justify its use in the ward environment or doctor's office. Patients being tested should consume a low peroxidase diet until the specificity of the new test has been fully evaluated.