The heterogeneity of patients’ responses in many treatment programs stems from the characteristics of their diverse background. Accordingly, paradigms of clinical decisions are shifting from “one-size-fits-all” rules to individualized treatment rules (ITRs). In statistical frameworks, an ITR is defined as a mapping of patient information to treatment recommendations, aiming to optimize the average future outcome of the program. In recent years, many researchers have developed methods for estimating optimal ITRs. This paper reviews some of the recent approaches, focusing on “indirect” (“-learning” and “A-learning”) and “direct” (“O-learning”) methods. We also briefly discuss further extensions of ITRs, which include multi-staged treatment rules also known as a dynamic treatment regime.