Two experiments explored the levels of text representation that mediate text repetition effects, following the Raney (2003) model. The magnitude of the repetition benefit in Experiment 1 supported predictions of Raney's model, indicating that the ease of forming a situation model contributed to the magnitude of the reprocessing benefit. In addition, representations organized around a good situation model were more sensitive to changes than were representations formed from reading without a good situation model. The results of Experiment 2 did not support the suggestion that the surface form and textbase are bound to a well-developed situation model, thereby limiting repetition effects to similar linguistic contexts. Rather, the nature of the repetition benefits in the present series of experiments are better explained by the degree o foverlap between passages at eachof the three levels of text representation.