Previous analyses of principal-agent problems with moral hazard assume the parties can commit to a contract that will not be renegotiated. We allow the contract to be renegotiated after the agent's choice of action and before the observation of the action's consequences. Here, the principal cannot induce the agent to take a high level of effort with probability one, since the principal would renegotiate the contract to shield the agent from risk, and in equilibrium the agent randomizes over effort levels. The optimal contract gives the agent a menu of compensation schemes: safe ones intended for low-effort workers, and risky ones for those whose effort is high. The optimal contract may give the agent a positive rent, in contrast to the case without renegotiation.