We propose a model of project selection and design of managerial compensation contract that features adverse selection and moral hazard. Our model generates the rather intuitive result that the ex ante probability of a specific project being selected (or, equivalently, its manager being hired) is increasing in the type of the project/manager. Ex post, however, the most capable manager (i.e., the one with the highest type) is not necessarily the one who will be hired to run a project. Basically, when the managers’ types are not identically distributed, picking the most capable manager or selecting the most promising project may actually be inconsistent with the provision of optimal incentives to alleviate the inherent agency problems. Therefore, our model offers a rational explanation to the phenomenon that apparently more capable candidates are occasionally passed over in recruitment and job promotion situations. Our analysis also holds obvious implications for firms’ capital budgeting decisions. If the severity of the principal-agent conflict is sufficiently great (say, between the headquater and the divisional manager) and if the verification of the true project type (the NPV value) by the headquarter is sufficiently costly, we may well see instances where corporate headquarters rationally allocate scarce resources to a lower-NPV project ahead of a higher-NPV project.