Five-year survival after resection of hepatic metastases from colorectal cancer is 25%. Although resection palliates some patients who do not live that long, 50% of patients so treated are not helped at all. Until ignorance of a cancer's real stage is resolved by improved techniques, the evaluation and choice of therapy can be based only upon knowledge of the natural history of untreated metastases and determinants of prognosis derived from treated patients. Analysis of the survival rates of 252 patients who had biopsy proven, unresected hepatic metastases that were the only evidence of residual disease shows the extent to which natural history, rather than resection, may determine length of survival-- and indicates the need for critical analysis of 2- and 3-year survival rates reported after any therapy. Study of 141 patients who had hepatic metastases resected shows that the stage of the primary lesion, being female, and the absence of extrahepatic metastases are significant determinants of favorable prognosis after resection of hepatic metastases.