Naive theories-knowledge structures with a causal or explanatory component-are examined in terms of initial development, resistance to change, and consequences. Three types of psychological processes are proposed to underlie the perseverance of naive theories. Illusory correlations occur when a biased sample of relevant events becomes distinctive by virtue of the numerical frequency of relevant events, the cognitive associations of relevant events, or the image-generating properties of relevant events. Data distortions occur when behavioral confirmation, biased attribution and recall, and biased assimilation processes operate. Use of some form of the availability heuristic underlies many of these effects. A model of naive theory perseverance and change is proposed. The model suggests where interventions might reduce potential biases that typically arise from use of naive theories. Evidence of successful use of several such debiasing techniques is presented.