Mixed methods were adopted to explore the effects of student-generated visualization on paper and on computers as a study strategy for middle school science concept learning. In a post-test-only-control-group design, scores were compared among a control-group (n=28), a group that was trained to visualize on paper (n=30), and a group that was trained to visualize on computers (n=34). The paper group and the computer group performed significantly better on the post-test than the control group. Visualization as a study strategy had positive effects whether students used paper or computers to generate their visualizations. However, no significant difference existed between the paper group and the computer group's scores. Qualitative results indicated that students who were trained to visualize on paper or on computers had positive attitudes toward the use of visualization as a study strategy, and they engaged more in purposeful tasks during study time than the control group. However, several participants in the computer group felt negatively about the use of visualization; they claimed that visualization demanded too much time and effort. Attitudes toward visualization as a study strategy differed within groups indicating that visualization might interact with learner characteristics. (Contains 2 tables.)