Learning from expository texts demands the processing of metatextual cues (rhetorical devices) and the activating of reading strategies. The main objective of this study was to examine whether profiting from written metatextual cues to launch reading strategies needs higher level of rhetorical competence than profiting from oral cues. Specifically, this study addresses two questions: (1) Is there a gap between the sensitivity to oral versus written metatextual cues depending on the student’s reading skill level? (2) Do the reader’s rhetorical competence, general reading comprehension, and decoding levels interact with the processing of each type of metatextual cue? Three hundred sixty-seven students (11–13 years old) summarized an expository text after reading it under one of the following four conditions: with written cues, with oral cues, with both cues combined, or with no cues. The less skilled readers who received oral or combined cues provided better summaries (they selected and organized the main ideas better) than the less skilled readers who received written cues or no cues. However, the performance of the more skilled readers was equal under the conditions with written cues, oral cues, and combined cues; these three groups outperformed the readers from the no-cues condition group. A multicategorical moderator analysis showed that following written cues demanded higher levels of general comprehension and rhetorical competence than following oral and combined cues. These data confirm that rhetorical competence is a specific capability for processing, especially written metatextual cues, and for overcoming the gap between the sensitivity to oral versus written cues.