Frail and physically or mentally disabled patients frequently have difficulty in tolerating formal colonic investigations. The aims of this study were to evaluate the accuracy of minimal-preparation CT in identifying colorectal carcinoma in this population and to determine the clinical indications and radiological signs with the highest yield for tumour. The CT technique involved helical acquisition (10-mm collimation, 1.5 pitch) following 2 days of preparation with oral contrast medium only. The outcome of 4 years of experience was retrospectively reviewed. The gold standards were pathological and cancer registration records, together with colonoscopy and barium enema when undertaken, with a minimum of 15 months follow-up. One thousand seventy-seven CT studies in 1031 patients (median age 80 years) were evaluated. CT correctly identified 83 of the 98 colorectal carcinomas in this group but missed 15 cases; sensitivity and specificity (with 95% confidence interval) 85% (78-92%) and 91% (90-93%), respectively. Multivariate analysis identified: (a) a palpable abdominal mass and anaemia to be the strongest clinical indications, particularly in combination (p<0.0025); and (b) lesion width and blurring of the serosal margin of lesions to be associated with tumours (p<0.0001). Computed tomography has a valuable role in the investigation of frail and otherwise disabled patients with symptoms suspicious for a colonic neoplasm. Although interpretation can be difficult, the technique is able to exclude malignancy with good accuracy.