This paper presents two cases of applying neural networks to extract knowledge for, subsequently, using it to support multicriteria contractor selection, in traditional design-bid-build projects with one-step selection processes. Different qualitative and quantitative selection criteria are taken into account, up to 22 and 9, respectively. The ? rst case includes a high number of input variables, making up a complex system related to complex and medium or large-sized projects. The second case is related to small projects in a medium-sized municipality. One advantage of these systems is that they can serve to ‘homogenize’ speci? c decision making in medium and large organizations. The paper also analyzes other pros of this approach, as well as the main problems.