We examine how stakes affect investor behaviors. In our unique setting, the same investors trade stocks in real accounts using their own money and, at the same time, trade in a simulated setting. Our real-world within-investor estimation produces strong evidence that investors exhibit stronger biases and perform worse in their higher-stakes real accounts than in their lower-stakes simulated accounts. Even with no monetary stakes, investors exhibit strong biases in their simulated accounts, and biases in the two types of accounts are strongly positively correlated. Such behavioral consistency suggests that low-stakes experimental methods, although imperfect, can be informative about real-world human behaviors. Using account data from two brokerage companies, we find that investors exhibit a stronger disposition effect on positions with greater portfolio weight. Hence, the finding that stakes-strengthening-biases may not be unique to the comparison between no-monetary and high-monetary stakes.