Colorectal cancer liver metastases (CRLM) are the main indication for liver resection in many western centers. Their management has drastically changed over the last two decades, mainly due to the increased safety of liver resections and the development of new chemotherapies. Previous contra-indications for a curative resection, such as extra-hepatic metastases are no longer the absolutes they once were, and we are permanently pushing the limits of curative treatment. As a consequence, the complexity of the management has increased in parallel, requiring a multi-disciplinary approach. This thesis presents the morbidity associated with liver resections, the evolution of the results of the resections for CRLM over time as well as their associated factors. The international results of the liver-first approach, a management of CRLM that has been designed at our institution, are presented. Finally, this thesis also presents the results of the resection of CRLM associated with simultaneous lung metastases, a condition previously thought to preclude curative treatment. No consensus exists for the management of the CRLM, mainly due to the multiple treatments that have been recently developed and to the lack of randomized clinical trials. Based on the current evidences in the literature and on our works presented in this thesis, an algorithm for management of CRLM is proposed. Evidences based on randomized trials are required to optimize our approach and to increase the curative and survival rates after liver resection.