Faecal alpha-1-antitrypsin and 51Cr-albumin losses in 42 patients with either gastrointestinal or hepatic disease were compared. The reference range was derived from measurements in 20 controls without gastrointestinal disease. Alpha-1-antitrypsin excretion was increased in patients with excessive 51Cr-albumin loss, and correlations were found between alpha-1-antitrypsin clearance and 51Cr-albumin excretion. Because of the considerable overlap of faecal alpha-1-antitrypsin excretion between controls and patients, sensitivity and specificity of the test were only 58% and 80%, respectively. This poor reliability could not be explained by sampling error or temporal variations in alpha-1-antitrypsin excretion. These results show that although faecal alpha-1-antitrypsin excretion correlates with 51Cr-albumin excretion when whole groups of patients are studied, its poor sensitivity makes it an unreliable measure of enteric protein loss.