The purpose of this research is to investigate the relative effectiveness of using three different question-prompt strategies on promoting metacognitive skills and performance in ill-structured problem solving by examining the interplay between peer interaction and cognitive scaffolding. An ill-structured problem-solving task was given to three groups. One group (Type QP) received instructor-generated question prompts that guided the problem-solving process; the second group (Type PQ) developed their own peer-generated questions; another group (Type PQ-R) developed their own question prompts first and revised them later with an instructor-generated question list. In this study, students in the QP group outperformed those in any other groups. The results revealed that providing instructor-generated question prompts was more effective than letting students develop their own questions, with or without revision, in ill-structured problem solving. Analysis of each of the four problem-solving stages revealed that the provided question prompts were more helpful in the stages of justification, and monitoring and evaluating than student-generated prompts. The difference between PQ and PQ-R groups is not statistically significant either overall or in any of the problem-solving stages.