In 3 experiments, we examined the effects of using concrete and/or abstract visual problem representations during instruction on students' problem-solving practice, near transfer, problem representations, and learning perceptions. In Experiments 1 and 2, novice students learned about electrical circuit analysis with an instructional program that included worked-out and practice problems represented with abstract (Group A), concrete (Group C), or abstract and concrete diagrams (Group AC), whereby the cover stories were abstract in Group A and concrete in Groups C and AC. Experiment 3 added a 4th condition (C-A) with a concrete cover story and abstract diagrams. Group AC outperformed Groups A and C on problem-solving practice in Experiments 1 and 2 and outperformed Group C on transfer across the 3 experiments; Group AC also outperformed Group C-A in Experiment 3. Further, Group A outperformed Group C on transfer in Experiments 2 and 3 and outperformed Group C-A in Experiment 3. Transfer scores were positively associated with the quality of the diagrams and the number of abstract representations drawn during the transfer test. Data on students' learning perceptions suggest that the advantage of Group AC relies on the combined cognitive support of both representations. Our studies indicate that problem solving is fostered when learners experience concrete visual representations that connect to their prior knowledge and are enabled to use abstract visual representations. (Contains 4 tables and 2 figures.)