This study reports the results of experiments using laboratory markets designed to test several predictions of economic behavior generated by DeAngelo's (1980, 1981) characterization of the auditor-client contractual relationship. These predictions address (1) auditor pricing behavior in the face of start-up costs and transactions costs incurred by the client when switching auditors and (2) the effect of start-up costs and transactions costs on auditor independence. The research was motivated by the recent concern expressed by policy-makers regarding the link between auditor pricing behavior and independence and the unobservability of auditor prices and independence in the naturally-occurring market. It was reasoned that, given the difficulty of obtaining data in the real world, laboratory market research would be a good "first step" in evaluating theories linking auditor pricing behavior to independence. The laboratory markets used to test the predictions were run in two stages. In the first stage, which used a sealed offer auction, subjects offered to sell an imaginary service to a computerized buyer in a series of five-auction markets, with certain payoffs. The contract prices in the first phase of the experiment were used to measure equilibrium predictions in conditions of certain payoffs. The second stage of the experiment, which tested predictions concerning independence and equilibrium prices in a world of uncertain payoffs, was identical to the first stage, except that payments were conditional on outcomes which depended on a decision rule specified by the sellers after each auction. The results of the experiments show significant decreases in independence as theoretical quasi-rents available to the incumbent seller increased. Furthermore, evidence on the relationship between lowballing and independence was inconclusive, due to the inequality of quasi-rents observed in cells where lowballing was allowed or restricted. With respect to pricing behavior, lowballing was observed in all markets where it was permitted, and increased as the theoretical quasi-rent stream increased. In addition, theoretical equilibrium price predictions under conditions of certain payoffs are weakly predictive of winning offers made by sellers. However, in settings with conditional payoffs, equilibrium predictions were not achieved in the experiment, perhaps due to design artifacts, or to the existence of an unspecified, alternative equilibrium.