Randomization and masking The randomization was to control or metastasectomy arms, with both arms similarly monitored. Patients were allocated equally, with stratification according to site. The sequence was generated at www.sealedenvelope.co.uk with minimization for age, sex, T(umour) stage, N(odal) stage, previous hepatic resection, interval since surgery for CRC, number of metastases and carcinogenic embryonic antigen (CEA) assay results, while retaining a random element.