The progression of colon cancer (CC) involves hematogenous and lymphatic spread to locoregional lymph nodes (LN), distant LN, and metastatic sites including the liver. The biological mechanisms that govern CC progression remain elusive. The Halsted model assumes an orderly, stepwise progression from the primary tumor to nearby nodes, henceforth to anatomically more distant nodes, and ultimately to distant organs. The Fisher model, on the other hand, regards the release of metastatic cells as early and essentially random events. The underlying biology has important implications for the ideal extent of surgery: when the Fisher model is correct, efforts to remove apical (central), extramesenteric, or para-aortic LN are unlikely to affect the oncological outcome. Recent data from phylogenetic studies suggest that cancer cell populations differ genetically among different LN stations and from distant metastases. Circulating tumor cells and other liquid biomarkers can be detected in the circulation of patients with early-stage disease. Local recurrence in CC is uncommon, and it is associated with a high risk of systemic progression and poor survival. Clinical studies comparing standard colectomy with extensive surgery (high ligation of the inferior mesenteric artery, complete mesocolic excision, D3 dissection, and para-aortic or extramesenteric node dissection) show that these techniques increase the LN count, while any beneficial effect on the risk of local recurrence or disease-free survival is at present uncertain due to the lack of controlled trials. Ongoing randomized trials comparing extensive vs. standard surgery for CC will generate important answers.