Generating written explanations is a popular learning strategy in an online learning environment. Students can explain to themselves (ie, self-explanations) or a peer-student (ie, instructional explanations). However, for improving learning from video lectures, it is unclear whether writing self-explanations is more beneficial than writing instructional explanations, and whether writing both types of explanation is more beneficial than writing only one type. We compared the learning-related outcomes of students who wrote explanations under one of four conditions: self-explanation (n = 30), instructional explanation (n = 30), self-explanation then instructional explanation (n = 30) and instructional explanation then self-explanation (n = 30). We assessed the participants' external and internal attention, explanation quality, and immediate and delayed learning performance. Students in the conditions that included self-explanations showed higher internal attention, as well as better immediate and delayed performance than those in the instructional explanations condition. In addition, students in the two combined conditions showed a higher level of organization and elaboration than those in the instructional explanations condition. These results suggest that students should write explanations to themselves while learning from video lectures.