The investigation of iron-deficiency anaemia (IDA) is a clinical problem which arises in virtually all branches of medicine. To audit the investigation of IDA, a computer-based laboratory record system was used to identify all women over 50 years of age and all men (n = 200) presenting to a single district laboratory with probable IDA in a six-month period. In 21 of 130 incident cases anaemia was clearly attributable to non-gastrointestinal disease. Of the remaining 109, 19% had investigation of both upper and lower gastrointestinal tract, 21% the upper gastrointestinal tract only, and 7% the lower gastrointestinal tract only. In 55 cases either no investigation was performed or only faecal occult blood tests. Eighteen months after presentation nine colorectal cancers, five gastric cancers and 11 peptic ulcers had been diagnosed; 21 patients had died, including two from colorectal cancers not detected when the IDA presented. This audit has revealed substantial underinvestigation of probable IDA, with serious but treatable conditions remaining undetected. Our findings, which we have no reason to believe are unrepresentative, indicate that policies are needed to ensure adequate investigation of IDA.