This paper addresses the methodology used to determine the amount of human resources needed to develop products. It is based on an empirical study of five firms in different industries. The methods rely on technical experts operating within budget constraints. The specific methods vary from relatively ad-hoc approaches to database-driven, centralized, and validated approaches. This variation is largely correlated with project and overall product development organization scale in a logical fashion in that larger projects and organizations move away from individual decision-makers toward centralized, objective, and validated methods. The project characteristics used by the experts and implicitly utilized in the database models to assess project complexity are also detailed in this paper. These characteristics are categorized based on a project complexity framework. We found that the scale of a project and the amount of "stretch" are the two most widely used characteristics for estimating project complexity. We found no utilization of the level of either component or task interactions in estimating project complexity by the firms studied. We also found no empirical evidence for interactions being a determinant of project difficulty despite a fair amount of theoretical concern with their importance in project complexity. In addition, we analyzed the accuracy of methods and conclude that since the accuracy at a higher level masks resource allocation problems at a lower level, the accuracy should be followed at multiple levels.