There is now a substantial body of literature analyzing pricing policies in unregulated markets when firms and consumers have asymmetric information about the quality of a product at the time the product is marketed. The existence of equilibria in which a variety of signaling mechanisms can be used to transfer a firm's information regarding its product to otherwise uninformed consumers is demonstrated in [1; 11; 14]. More recently, Bagwell and Riordan analyzed the learning procedure of consumers when a product of unknown quality is introduced to the market, and some fraction of consumers becomes informed over time [9]. In these models, the firm has prior knowledge of the exact quality, and consumers either correctly infer quality prior to purchase, because of the incentive structure of the equilibrium signaling scheme, or learn quality immediately upon purchase.' The extensive literature on statistical testing and quality control, as well as the frequent news items on product recalls, suggests that, prior to observation of market performance, a firm's knowledge of the properties of its product may not be exact.2 There is another body of literature which examines the effectiveness of various forms of liability schemes on the level of care and the severity of the warnings firms choose when producing and marketing their products. Models by Marino, Shavell, and Cooter have the assumption that the market will not provide the firm with sufficient incentives to provide the socially efficient levels of care and warnings, and these market incentives must be supplemented by liability rules of various kinds [10; 13; 4]. The particular liability rule which is chosen varies with the assumptions of the model; Marino showed that the existence of scale effects in consumption could alter the common conclusions regarding the relative effectiveness of strict liability and regulation. This paper examines the incentives a firm faces to recall a product which is discovered to be