The "illusory correlation" effect occurs when perceivers make memory-based judgments about two unequal-sized groups of people after receiving information about their behaviors. The stimulus information includes equal proportions of positive and negative behaviors by members of each group, but perceivers generally perceive an association between membership in the smaller group and the less frequent type of behavior. A quantitative model of long-term memory, implemented as a computer simulation, reproduces several known properties of the effect. The simulation uses D. L. Hintzman's memory model, which involves the storage and retrieval of specific experiences but no differential attention or distinctiveness-based encoding or retrieval biases. Besides generating the basic effect, the model also reproduces several detailed patterns of results that have been interpreted as supporting the traditional account for the phenomenon in terms of distinctiveness-biased encoding. This demonstration that there is more than one viable theoretical explanation for the effect has seveval implications for future research and theory on illusory correlation. An Appendix briefly describes failures to replicate the same effect using several PDP (distributed memory) models. © 1991.