Background: Previous studies have shown that undergraduates improve their answering and monitoring accuracy when they exclusively practice and expect inferential questions after reading. This study examined whether children with poor comprehension, who struggle particularly with inferential questions, would benefit from similar practice with and without feedback. Methods: To address this question, 44 poor comprehenders and 44 control participants from 6th-9th grades practiced answering literal or inferential questions after reading each of three texts. They were also asked to predict their success in these questions, whereas some received feedback on their prediction (monitoring) accuracy. Then, participants read an additional three texts, but answered both practiced and unpracticed types of questions after reading all texts. They also predicted their success after reading each text. Results: Both poor and good comprehenders answered literal questions more accurately when they had practiced. However, only good comprehenders improved their answering of inferential questions when they had practiced. No differences were found between the groups in monitoring accuracy. Feedback had a positive effect on answering accuracy, irrespective of practice. Conclusions: Poor comprehenders differentiate to some extent between literal and inferential questions and are flexible enough to execute a different text processing plan for each type of questions. However, they presumably lack the knowledge and/or resources to execute inferential processing efficiently during reading. Moreover, all children seem to have difficulty with comprehension monitoring. Practicing and/or expecting one type of questions, with or without feedback, is insufficient for improving this ability.