When investment opportunities arrive one at a time and are reviewed sequentially, a corporation's optimal policy differs from a standard net present value rule if the corporation exercises control over an industry state variable and control is costly. The first condition presupposes a degree of market power for the firm; the second occurs if corporate investment decisions are imperfectly reversible.

To address the problem of optimal investment in this context, a firm's investment decisions are modeled as a Markov reward process. The causes of economic irreversibility are discussed and general propositions concerning the optimal investment policy are derived. These propositions are then applied to the optimization of an exploration program by an oligopolistic firm (a price leader). Under particular demand and distributional assumptions, solutions for the optimal decision rule and the value of the exploration program are obtained and their properties examined.