This paper explains why people value self-confidence, and how this concern shapes their informational strategies and intertemporal decisions. The theory has applications in areas as diverse as labour supply, savings and investment, or education and career decisions. People generally have imperfect knowledge about their abilities, which in most tasks are complementary to effort. Self-confidence thus enhances motivation, and this gives a time--inconsistent individual a strong incentive to build up the self-esteem of his future selves, so as to limit their procrastination. The benefits of confidence-maintenance must, however, be traded off against the risks of overconfidence. Moreover, rational inference implies that the individual cannot systematically fool himself. The model explains why people often choose to remain ignorant about their true abilities, or 'blind' to important signals from their work, personal or market environment; and why they sometimes deliberately impair their own performance or choose overambitious tasks in which they are sure to fail (self-handicapping). It also provides a formal account of (endogenously) selective memory or awareness, such as the tendency to remember one's successes more than one's failures. This result, in turn, helps explain why most people have overoptimistic assessments of their own abilities and accomplishments (self-serving beliefs). Another important result is that this 'psychological immune system' can lead to multiple intrapersonal equilibria in cognitive strategies, self confidence, and behaviour. Moreover, while 'positive thinking' and similar forms of self--deception can improve ex-ante welfare, they can also be self-defeating.