The project management literature argues that most projects fail, and yet, paradoxically, increasing numbers of proposals for new initiatives attract funds. In order to resolve an apparent 'investment-in-failure' paradox, this paper questions the methodology used in the literature to judge project performance and to decide on funding new projects. Using results from a field study, the authors describe a project performance framework that both expands and extends traditional approaches. They argue that the conventional test of project performance is not only fundamentally flawed, but also irrelevant to decision-makers. In response, drawing on 'principal-agent', 'regret' and 'contingency' theories, the authors propose a new methodology to assess projects based on the concept of 'worth'. According to this approach, performance is judged at three separate levels: project management, project ownership and project investment. These three tests allow distinct judgements to be made about the respective performances of the project manager, the project owner and the investment represented by the original funding decision. To the extent that financial crises are associated with project failure, such a framework may prove useful, because it would support better investment decision-making.