AimsWe tested whether combining constructive learning prompts with retrieval practice (constructive retrieval) positively affects delayed learning outcomes (factual knowledge, comprehension), metacognitive accuracy, and motivation. Sample & methodsUndergraduate students (N = 152) learned from a video-recorded lecture, and then engaged in follow-up learning activities involving retrieval practice, restudy, elaboration, or a combination of retrieval practice with elaboration. We ensured that initial retrieval success was high and that all groups had the same expectations about the nature of content to be tested. We assessed both factual knowledge and comprehension, as well as the respective indicators of metacognitive-monitoring accuracy (metamemory, metacomprehension). We assessed self-efficacy and situational interest as motivational variables. ResultsIn a one-week-delayed posttest, elaboration improved factual knowledge and comprehension. This pattern held whether learners just elaborated or additionally engaged in retrieval practice. Having students self-generate their own examples during retrieval practice improved comprehension, but merely letting students retrieve experimenter-provided examples did not. Students' self-reported mental effort mediated the factual knowledge benefits of retrieval practice. The number of students’ self-generated elaborations mediated comprehension benefits. Students engaging in constructive retrieval while learning revealed the highest metacognitive-monitoring accuracy in factual knowledge and comprehension. We found no differences between conditions regarding situational interest and self-efficacy. Motivation to reuse the respective learning strategies was higher in all conditions compared to restudy. ConclusionConstructive retrieval seems advantageous when considering its benefits on learning outcomes and metacognition. When working on retrieval tasks, students only profited from self-generated, but not from provided, examples with respect to comprehension.