Research indicates that the combination of correct and erroneous examples has a positive effect on knowledge acquisition and transfer if learners have adequate prior knowledge. If only erroneous examples were provided, students with low prior knowledge may need additional supports so that their procedural skills can be effectively promoted. For this reason, this study, taking subtraction as the learning domain, conducted a 2 × 2 factorial design of quasi-experiment varying in example types (correct vs. erroneous examples) and feedback settings (with vs. without feedback). Results indicated that feedback may promote transfer and retention in learning from worked-out examples. However, there was no significant difference between correct and erroneous examples in helping students learn subtraction procedures. According to the results of our questionnaire on cognitive load, it was likely caused by the complexity of the organization and presentation of erroneous examples. We suggest that clear and intelligible presentation can be supported by computer technologies.