Intention-to-treat analysis is a technique used in randomized controlled trials (RCTs), where patients are compared--in terms of their final results--within the groups to which they were initially randomized, independently of receiving the allocated treatment, having dropped out of the study or having violated the initial protocol (for whatever reason). In other words, it constitutes an analysis of the results based on the treatment arm to which the patients belong due to the initial random allocation, and not on the treatment actually received (active or placebo). Intention-to-treat analysis permits the pragmatic evaluation of the benefit of a treatment change, and not the potential benefit in patients getting the pre-planned allocated treatment only. Full application of the intention-to-treat principle is only possible in those circumstances where all results from all patients are available. A significant number of RCTs state that they use intention-to-treat analysis, but the ways that violations of randomized allocation is handled varies considerably. Many trials present flaws in gathering primary data, and their methods of handling this problem are normally poor. Lastly, the intention-to-treat principle is frequently poorly described and applied. In this article we will present the importance, utilization, limitations and errors of intention-to-treat analysis.