This paper analyzes revisitations in online games focusing on two types of revisitations: game revisitations and area revisitations. A player revisits a game and areas therein with purposes. For game revisitations, we conduct a large-scale analysis using Shen Zhou Online access log collected for nearly 6 years consisting of 50,000 characters and have succeeded in using in the information on game revisitations, together with the login time and login frequency information, for predicting the players who will be absent from the game. For area revisitations, we conduct yet another large-scale analysis using World of Warcraft access log collected for 2 years consisting of more than 60,000 characters and have discovered four main groups of area revisitation patterns. We also discuss in the paper how our findings can be utilized to support both game developers and players.