Readers' memory for belief-consistent texts is often stronger than for belief-inconsistent texts (text-belief consistency effect). However, presenting belief-consistent and belief-inconsistent texts alternatingly reduces the discrepancy between the memory strengths of belief-consistent and belief-inconsistent texts. The present study used eye tracking to examine the cognitive processes underlying the text-belief consistency effect and how it is moderated by the mode of presentation. At 2 points of measurement, 41 university students read 2 belief-consistent and 2 belief-inconsistent texts on 2 scientific issues blocked or alternatingly. Comprehension outcomes were assessed with an essay task. First-pass rereading times were longer for belief-inconsistent information for participants with strong beliefs. A blocked presentation increased this effect and yielded longer first-pass rereading times for belief-inconsistent claims and a text-belief consistency effect in the essay task. An alternating presentation increased immediate and delayed processing of belief-inconsistent information and reduced the text-belief consistency effect, especially in readers making more lookbacks.