A clinical trial to determine the antitumor activity of recombinant interferon-gamma (rIFN-gamma) was conducted in 36 patients with advanced colorectal cancer. Severe constitutional symptoms were seen in the first five patients who received rIFN-gamma as a 2- to 4-hour intravenous infusion, and this method of administration was therefore abandoned. One transient partial tumor regression was observed in the 31 patients who received treatment by the intramuscular route of administration. Although a clinically tolerable regimen suitable for outpatient administration was developed, rIFN-gamma given in this dose and schedule had minimal antitumor effect for the treatment of advanced colorectal cancer. NK activity was depressed within 48 hours of rIFN-gamma administration but became significantly higher than normal controls or pretreatment levels during therapy despite disease progression, indicating discordance between augmentation of this immune parameter and tumor status. Serum CA 19-9 levels were unusually high and increased significantly more than CA 125 or CEA during rIFN-gamma treatment. This observation suggests that rIFN-gamma could augment localization of anti-CA 19-9 used to diagnostically image or therapeutically target a labeled chemotherapy agent to tumor in those patients expressing this antigen.