Short-run maximization of firm profits provides the sole justification for transfers at marginal cost. This conclusion is based, however, on a theory of the firm that ignores precisely those information and agency costs that make transfer pricing necessary. An alternative approach is to replicate free market forces within the firm, allowing buying and selling divisions complete freedom to negotiate prices. We refer to this as laissez-faire transfer pricing. While short-run profits are suboptimized, this approach provides superior measures of divisional performance, reduces many of the agency costs associated with marginal cost pricing, and is more likely to lead to the maximization of long-run firm profits. In this paper, a neoclassical model of the firm is used to compare firm and divisional profits under both marginal-cost and laissez- faire transfer pricing. The differences in the level and allocation of these profits under the two systems quantifies the tradeoff between profit losses due to sub-optimization costs and improved performance measurement associated with deviating from the marginal cost rule. Our results indicate that transferring above marginal cost will significantly reduce total firm profits only when there are few outside customers for the transferred product. However, these results also indicate that when there are few outside customers, transfers at marginal cost will seriously distort divisional profit allocations. As a result, the firm faces a serious dilemma. It can maximize short-run profits, seriously distort divisional profit measurement and incentives, and, consequently, increase the probability that long-run profits will fall. Alternatively, it can sacrifice some short-term profits, obtain superior measures of divisional contributions to firm profits, and improve the chances of maximizing shareholder wealth in the long-term.