Within a narrow time frame after acute myocardial infarction (AMI), spontaneous or therapeutic reperfusion can lead to myocardial salvage, preservation of left ventricular (LV) function and improved survival. However, several studies have shown that the improvement in prognosis of patients with sustained patency of the infarct-related artery (IRA) is out of proportion to the observed improvement in ventricular function. Also, patients who undergo late thrombolysis (i.e. after the traditional 4- to 6-hour time window for myocardial salvage), show improvements in LV function and prognosis. These observations suggest that other mechanisms contribute to the beneficial effects of reperfusion in addition to the salvage of ischaemic myocardium. Experimental and clinical data support the concept the late reperfusion prevents LV dilatation and improves LV function by limiting infarct expansion and ventricular remodelling. These benefits are independent of any limitation on infarct size. Ventricular remodelling is a dynamic process, starting in the acute phase with infarct expansion, and progressing to LV dilatation and hypertrophy. The remodelling process has regional and global effects on wall thickness, and on chamber size, shape and function. Possible mechanisms by which late reperfusion limits infarct expansion independently of infarct size reduction include accelerated scar formation and healing, increased stiffness of the myocardium due to a blood-filled coronary vascular bed, and preservation of a subepicardial rim of viable cells. Evidence from trials of thrombolytic therapy given both early and late after AMI indicates that patency of the IRA is a critical predictor of progressive remodelling and LV dilatation in the long term. Long-term follow-up studies also show the importance of residual stenosis or reocclusion in this context. Thus, adequacy of reperfusion of the infarct bed may be more important than timing of reperfusion in attenuating LV dilatation. Since the pathogenesis of remodelling is influenced by many factors, the optimal therapeutic approach may involve a combination of treatment modalities targeted at several of the underlying processes. In fact, a variety of interventions show great promise in their ability to influence the remodelling process. In particular, there is increasing evidence that remodelling can be prevented through the use of late reperfusion therapy and therapy aimed at maintaining IRA patency in the long term. Based on current evidence, a rational strategy for the primary prevention of ventricular remodelling should involve intervention starting very early after infarction, spanning the entire healing process and continuing well beyond. Prospective clinical trials are required to evaluate late reperfusion therapy, as well as strategies for reducing residual stenosis, preventing reocclusion, and maintaining IRA patency in the long term.