In this study, the effects of reflective self-explanations on conceptual understanding and problem solving are investigated in a domain of computer programming. After completing a multiple-choice test, 33 students were asked to reflect on and explain why their answers were correct or incorrect. Groups were divided for post hoc analysis based on the degree of reflective self-explanation and prior knowledge. Results indicated that students who actively engaged in the reflective self-explanation process performed better in problem solving tasks with minimal guidance. Prior knowledge was critical for effective reflection, especially recalling principles. Although low prior knowledge students tended to persist with their faulty mental models, self-explanation activities helped them to repair their models. The relationship between reflective self-explanation and prior knowledge is discussed. (Contains 4 tables and 4 figures.)