There is a sense in which simple interference effects are destroyed when a system's environment becomes correlated with its state. This phenomenon is called decoherence. I shall consider three approaches here. According to one, decoherence alone explains why we get determinate records. According to another, decoherence helps one to formulate a satisfactory interpretation of Everett by selecting a globally preferred basis that makes the right physical facts determinate in each Everett branch. According to a third, decoherence sects a locally preferred basis for each observer that makes the right physical facts determinate from the perspective of that particular observer.