This study was performed prospectively to assess the effect of systemic chemotherapy (FOLFIRI protocol) in patients with initially unresectable colorectal liver metastases (CRLM) and, after performing liver resection in patients with downsized metastases, to compare the postoperative and long-term results with those of patients with primarily resectable CRLM. Records from a prospective database including all consecutive admissions for CRLM between June 2000 and June 2004 were reviewed. The analysis addressed all patients who underwent hepatectomy for primarily resectable CRLM (Group A), or underwent chemotherapy for primarily unresectable CRLM and among these, particularly the patients who were finally resected after downsizing of CRLM (Group B). There were 60 primarily resected patients (Group A). Forty-two other patients underwent chemotherapy; after an average of nine courses, 18 of them (42.8%) with significantly downsized lesions were explored and 15 (35.7%, Group B) were resected, whereas three had peritoneal metastases. Group B differed from Group A for a significantly higher rate of synchronous CRLM upon diagnosis of colorectal cancer, a larger size of CRLM upon evaluation in our center, and a lower rate of major hepatectomies (20.0% vs. 51.6 %) at surgery. No patient in Group B had positive margins of resection. Operative mortality was nil and morbidity was 20.0% in both groups. In Group B vs. Group A median survival after hepatectomy was 46 vs. 47 months (n.s), 3-year survival rate was 73% vs. 71% (n.s.), disease-free survival rate was 31% vs. 58% (p = 0.04) and, at a median follow-up of 34 months, tumor recurrence rate was 53.3% vs. 28.3% (n.s.). Four out of the eight Group B patients with recurrence underwent a re-resection, and were alive at 9 to 67 months after the first resection. These results show that in about one-third of the patients with primarily unresectable CRLM, downsizing of the lesions by chemotherapy (FOLFIRI protocol) permitted a subsequent curative resection. In these patients, operative risk and survival did not differ from the figures observed in primarily resectable patients and, in spite of a lower disease-free survival with more frequent recurrence, re-resection still represented a valid option to continue treatment.