Theory-building is the engine of the scientific enterprise and it entails (1) the generation of new hypotheses, (2) their justification, and (3) their selection, as well as collecting data. The orthodox views maintain that there is a clear logical and temporal order, and distinction, between these three stages. As a matter of fact, not only is this tenet defective, but also there is no way to solve these three issues in the way advocated by traditional philosophy of science. In effect, what philosophy of science tells us is that (a) there is not an infallible logic, in the sense of a simple set of logical rules, to justify and confirm a hypothesis, and (b) the process of generation of hypotheses is not unfathomable, but can be rationally investigated, learned and transmitted. So, as an alterative, I discuss the heuristic approach to theory-building, especially the one based on problems, and I argue that it offers a better way of accounting for theory-building than the traditional ways.