This paper reformulates the continuous-time principal-agent model of Holmstrom and Milgrom by adding a dynamic incentive compatibility constraint. The original Holmstrom-Milgrom solution applies under this dynamic constraint, but not generally under the static constraint formulated by Holmstrom and Milgrom. With only a static initial incentive compatibility constraint, the Holmstrom- Milgrom solution is an equilibrium for, essentially, only a special quadratic cost-of- effort function._x000B__x000B_