This study examines the accuracy of colorectal cancer radioimmunodetection. Twenty-seven patients with a history of histologically-confirmed colonic or rectal carcinoma received a high-titer, purified goat anti-CEA IgG labelled with 131-I at a total dose of at least 1.0 muCi. Various body views were scanned at 24 and 48 hours after administration of the radioantibody. Three additional cases were evaluated; one had a villous adenoma in the rectum and received the 131-I-labelled anti-CEA IgG, while two colonic carcinoma patients received normal goat IgG labelled with 131-I. All of the 7 cases with primary colorectal cancer showed true-positive tumor localization, while 20 of 25 sites of metastatic colorectal cancer detected by immune scintigraphy were corroborated by other detection measures. The sensitivity of the radioimmunodetection of colorectal cancers (primary and metastatic) was found to be 90% (true-positive rate), the putative specificity (true-negative rate) was 94%, and the appraent overall accuracy of the technique was 93%. Neither the case of a villous adenoma receiving the anti-CEA IgG nor the two cases of colonic cancer receiving normal goat IgG showed tumor radiolocalization. Very high circulating CEA titers did not appear to hinder successful tumor radiolocalization. These findings suggest that in colorectal cancers the method of CEA radioimmunodetection may be of value in preoperatively determining the location and extent of disease, in assessing possible recurrence or spread postoperatively, and in localizing the source of CEA production in patients with rising or elevated CEA titers. An ancilliary benefit could be a more tumor-specific detection test for confirming the findings of other, more conventional diagnostic measures.