Interest in knowledge management for organizations stems from the recognition that knowledge is an essential resource of a firm for sustainable competitiveness ftom a resource-based view of strategy and competition. Under this lens, knowledge is treated as a unique resource of a firm that is not easily imitated by competitors. A major reason for inimitability is causal ambiguity. Unfortunately, while causal ambiguity is useful in hindering undesirable transfer of knowledge across firm boundaries, it also hinders desirable transfer of knowledge within organizations. A large number of firms have seen dismal results from investments in knowledge management. One overlooked and under-researched perspective is the effect of the causal ambiguity as it applies to knowledge, otherwise known as knowledge ambiguity. The purpose of this paper is to highlight the theoretical relationship between knowledge ambiguity and knowledge management practices. To do so, the paper extends a model proposed by March in his 1991 paper. Two types of knowledge ambiguities are added to the model: first, knowledge that brings about good performance is not always good, and second, knowledge that leads to poor performance is not always bad knowledge. False positives ambiguity is a lack of ability to quantify the former relationship, and false negatives ambiguity refers to the latter. Second, the employee and the organization update knowledge at different rates. The analysis reveals that under false positives ambiguity, knowledge should be carefully validated, whereas under false negatives ambiguity, knowledge leading to both good and poor performance must always be collected. A result that is not as intuitive is that under false negatives ambiguity, the organization needs to exert effort to obtain the correct knowledge, whereas under false positive ambiguity, the employee needs to exert effort to obtain correct knowledge. These results highlight the differences in organizing the knowledge management function: when is a knowledge manager useful and how much discretion should such a manager have in a knowledge repository.