We analyze the optimal combination of promotion tournaments and individual performance pay in an employment relationship. An agent's effort is non-observable and he has private information about his suitability for promotion. We find that the principal does not provide individual incentives if it is sufficiently important to promote the most suitable candidate. Thus, we give a possible explanation for why individual performance schemes are less often observed in practice than predicted by theory. Furthermore, optimally trading off incentive and selection issues causes a form of the Peter Principle: The less suitable agent has an inefficiently high probability of promotion.