BACKGROUND: The outcome of liver resection for colorectal liver metastases (CRLM) appears to be improving despite the fact that surgery is offered to patients with more-severe disease. To quantify this assumption and to understand its causes we analyzed a series of patients on the basis of a standardized severity score and changes in management occurring over the years. METHODS: Patients' characteristics, operative data, chemotherapies and follow-up were recorded. CRLM severity was quantified according to Fong's clinical risk score (CRS), modified to take into account the presence of bilateral liver metastases. Three periods were analyzed, in which different indications, surgical strategies and uses of chemotherapy were applied: 1984-1992, 1993-1998, and 1999-2005. RESULTS: Between January 1984 and December 2005, 210 liver resections were performed in 180 patients (1984-1992, 43 patients; 1993-1998, 42 patients; 1999-2005, 95 patients). CRLM severity increased throughout the time periods, as did the use of neoadjuvant chemotherapies, repeat resections, and multistep procedures. While the disease-free survival did not improve over time, the 1-, 3- and 5-year overall survival rate increased from 85%, 30%, and 23% in the first period, to 88%, 60%, and 34% in the second period, and to 94%, 69%, and 46% in the third period. CONCLUSIONS: Analysis according to the CRS showed that despite the fact that patients had more severe disease, the overall survival improved over the years, mainly thanks to more aggressive treatment of recurrent disease. Management of advanced CRLM should, from the start, take into account the likelihood of secondary procedures.