In a multicentre prospective trial, 281 patients undergoing "curative" resection for colorectal cancer were followed for a mean of 13.6 months (SD 7.2 months). The presence or absence of sialomucin at the resection margin was studied histochemically using the high iron diamine-alcian blue (HID-AB) stain. There were 49 deaths relating to tumour recurrence: 21 in the sialomucin positive group (n = 77) and 28 in the sialomucin negative group (n = 204) (p less than 0.02). Life table survival was correlated against the presence or absence of sialomucin in the resection margin. At the mean follow-up (13.6 months) 85.6% of patients were alive in the sialomucin negative group, and 76.4% of patients were alive in the sialomucin positive group. Regression analysis predicts 32.8% and 18.9% five year survivals for sialomucin negative and positive groups respectively. There was no significant statistical correlation between the presence of sialomucin in the resection margin and the Dukes staging, site or tumour differentiation. The appearance of sialomucin in either resection margin appears to be an early marker of poor prognosis for patients with colorectal cancer.