Test-potentiated learning occurs when testing renders a subsequent study period more effective than it would have been without an intervening test. We examined whether testing only a subset of material from a multimedia lesson would potentiate the restudy of both tested and untested material. In Experiments 1a and 1b, participants studied a multimedia lesson about star formation and then read facts (read participants) or took a test (test participants) on half the information from the lesson. All participants then restudied the entire lesson before taking a final cued-recall test after a 5-min (Experiment 1a) or 20-min (Experiment 1b) delay. On the final test, evidence for test potentiation was observed: Test participants recalled just as much untested material as tested material, whereas read participants recalled less of the unread information than the read information. Furthermore, untested information was better recalled than unread information. The results from these experiments suggest that taking a test on some information from a multimedia lesson aids restudy of both tested and untested information more effectively than does reading that same subset of information. Additionally, in Experiment 2, we found the potentiating benefit of testing to extend to the learning of new, unrelated information--specifically, the learning of a multimedia lesson on a completely different topic. Across all experiments, however, the test participants--despite learning more as reflected by final test performance--reported lower confidence regarding how well they had learned the material than did the read participants.