Previous research has shown that observation can be effective for learning in various domains, such as writing, reading and creative art work. By observing models at work, students can develop strategic knowledge and they might also change their conception of what the modeled skill involves. The question in this study is which instructional approach is more effective for students' processes of studying text: learning by observation or learning through practice? We designed an intervention that enables students to learn from observation of video models. The models in these videos are peers, who read and learn a history text while thinking aloud. In a pre-test/post-test control group design we assigned 52 Dutch students (15-18 years old) to one of two conditions; one observational learning condition and a control condition (learning through practice). In the observational learning condition, students were asked to observe and evaluate the thinking processes of two peers on video, and decide which was the weaker and which was the stronger better performer of a study task. In the control condition, students received direct instruction and practice in reading and learning strategies. We measured students' attitude and self-reported use of learning strategies with a questionnaire (at pre- and post-test) and a learner report (post-test only). Students' use of strategies at pre- and post-test was measured with a think aloud task with eight participants of each condition. This showed that students in the experimental condition checked their own understanding of the text while studying for a history test more often than students in the control condition. Moreover, students in the experimental condition reported more learning experiences in their learner reports than those in the control group. In these reports, this group also reported almost 10% more metacognitive learner experiences.