This study tested an instructional design to improve students' synthesis performance in a specific academic subject, Science Orientation, which aimed to teach students how to critically evaluate scientific debates. The design included three components: 1) students construct a task definition via a learning strategy based on comparing and contrasting texts and processes, 2) students comprehend source information via a read-stop-think-note strategy, and 3) students connect source information critically via a semantic-textual transformation strategy.After several design iterations, the instructional design was tested in a quasi-experimental experiment with a pretest-posttest. Seven 10th grade classes participated in the intervention (n=129), four in the control condition (n=86). The design seemed feasible for teachers, students completed most learning tasks as intended and evaluated the course positively. Furthermore, texts written in the experimental condition at posttest were rated significantly higher than those written in the control condition on the instructed aspects: representation of source information, intertextual integration, and critical stance. This instructional design appears to have potential for helping students improve their comprehension of scientific debates and comprehensive writing. In the discussion we propose that the instructional design might be a general format for learning to synthesize domain specific information from contrasting sources.