Over the past two decades, the perennial low success rates of elementary students in math problem-solving and the difficulties experienced by teachers in helping their students with this type of task has become quite a hot topic. In response, several instructional interventions aiming to develop an expert and reflexive approach to problem-solving have been designed. However, these interventions are based on two contrasting learning approaches, either teaching the components of the problem-solving process at the same time or teaching them one at the time. The two approaches have never been compared. Moreover, they have mainly been assessed in terms of cognitive outcomes. Yet, recent studies stress the importance of analyzing the cognitive, motivational and emotional processes involved in problem-solving learning together in order to gain a full understanding of the process. Addressing these limitations is essential to enhance our understanding of problem-solving learning and to design more effective interventions. This paper focuses on this issue by investigating whether it is preferable as regards cognitive, motivational and emotional outcomes, to teach the problem-solving process in all its complexity or one component at a time. This issue is handled both for novice and expert solvers. Data were gathered among 267 upper elementary students. Findings showed that both learning approaches support the short- and long-term acquisition of cognitive problem-solving strategies, regardless of the student’s profile. However, beneficial emotional and motivational outcomes occur only when the problem-solving process is taught in all its complexity, i.e., makes sense for the learner. Novice solvers made less use of the help seeking strategy and persisted more.