We examine the role of security design when lenders make inefficient accept or reject decisions after screening projects. Lenders may be either “too conservative,” in which case they reject positive-NPV projects, or “too aggressive,” in which case they accept negative-NPV projects. In the first case, the uniquely optimal security is debt. In the second case, it is levered equity. In equilibrium, profitable projects that are relatively likely to break even are financed with debt, while less profitable projects are financed with equity. Highly profitable projects are financed by uninformed arm's-length lenders.