5-FU, semustine (MeCCNU), triazinate (TZT), and razoxane (ICRF-159) have each shown activity against advanced colorectal cancer in studies by at least two investigative groups. Objective response rates, however, have been low, without evidence of increased patient survival. The hope of this study was that enhanced activity might result from giving these agents in two-drug combinations. There were 167 eligible and evaluable patients randomized among the programs: 5-FU at a dose of 500 mg/m2/day by iv push X 5 (F); 5-FU at a dose of 400 mg/m2/day iv X 5 plus TZT at a dose of 175 mg/m2/day iv X 3 (FT); 5-FU at a dose of 400 mg/m2/day plus ICRF-159 at a dose of 600 mg/m2/day orally X 3 (FI); MeCCNU at a dose of 150 mg/m2/day orally plus TZT at a dose of 200 mg/m2/day iv X 3 (MT); MeCCNU at a dose of 150 mg/m2 orally plus ICRF-159 at a dose of 500 mg/m2/day orally X 3 (MI); and ICRF-159 at a dose of 425 mg/m2/day orally X 3 plus TZT at a dose of 125 mg/m2/day iv X 3 (IT). Patients with limiting conditions (serum creatinine greater than 1.5 mg/dl or elevated bilirubin) were randomized among programs F, FI, and MI. Objective response rates by treatment arm were: F--13% (four of 31 patients); FT--13% (four of 31); FI--15% (four of 27); MT--11% (three of 28); MI--13% (four of 32); and IT--6% (one of 17). Response rates of combination arms were not significantly larger than those of 5-FU alone. With regard to survival, patients initially treated with 5-FU alone had the most favorable experience (median, 10.8 mos). Multivariate analysis showed the following factors to have a significant and independent influence on survival: Eastern Cooperative Oncology Group performance score, grade, site of indicator lesion, and the presence of 5-FU in the treatment regimen. Toxic effects most frequently seen were nausea, vomiting, thrombocytopenia, leukopenia, diarrhea, stomatitis, alopecia, and dermatitis. The incidence and severity of toxicity were roughly comparable among the six treatment arms.