Following resection of their tumour, 162 patients with colorectal cancer entered a prospective randomized controlled clinical trial of adjuvant oral razoxane. Thirty-one patients were Duke's group A; 49 group B; 61 group C; and 17 group D; an additional four patients were randomized in error. The adjuvant group received the usual clinical care and 125 mg razoxane twice daily for 5 consecutive days (monday-friday) every week indefinitely. Control patients received the same clinical care as the adjuvant group, but no razoxane. At 3 years, 134 patients (84%) are evaluable. The recurrence rate in the first 6 months was 20% and 28% respectively in the Duke's B and C controls compared with 4% and 9% in the corresponding razoxane treated patients. Most recurrences occurred within the first 6 months from randomization. When all patients as randomized are included in the analysis of survival, irrespective of whether they were cured by surgery (Duke's A), had advanced cancer (Duke's D), or took no razoxane when randomized to take it, then as might be expected any differences there may be between the razoxane-treated and control patients with minimal residual disease (Duke's B and C) are so distorted that the p value of the difference in survival was 0.93. If however only patients with Duke's group B or C are taken (49 controls and 47 treated), log-rank analysis reveals a difference in the cancer mortality curves (p = 0.07). If patients who had been randomized to take razoxane, but who had not taken it at any time (and therefore received the same treatment as controls) are analysed with the controls, the difference between the two groups increases further, with p less than 0.05. The razoxane-treated patients experienced no significant toxicity apart from a readily reversible mild leukopenia in 52% while gastrointestinal symptoms necessitated stopping the drug in only four patients. These four all took the drug for less than 4 weeks. Because there was no toxicity to subtract from any benefit razoxane adjuvant treatment produced and the quality of life was not impaired, the therapeutic benefit of surgery wsa increased to the extent that razoxane increased survival of patients with Duke's B and C tumours.