Elaborative feedback (EF) containing explanations on students' responses benefits learning. Computer-based environments provide learners with EF in different ways, for example, on an immediate question-by-question basis or after answering a set of questions. Recent findings also suggest that delaying EF enhances learning. However, it is unclear to what extent different types of delayed-EF favour students' performance. This study examines whether and how two types of delayed-EF (question-based vs. summative) influence students' question-answering performance and final learning over immediate-EF. One hundred thirty-three secondary-school students read a scientific text and answered 12 multiple-choice questions in a computer-based environment. A day later, students completed a final learning test with 20 open-ended questions. Results showed that neither question-based delayed EF nor summative delayed EF outperformed immediate EF. However, EF moderated the relationship between students' prior knowledge and their performance outcomes, suggesting that students with higher levels of prior knowledge receiving summative delayed EF benefited more.