This thesis is concerned with the concept of a “ good enough” requirements document. It takes the position, based on empirical observations, that standard prescriptive approaches have failed to identify the necessary and sufficient characteristics of a good requirements document, because what is good enough in one situation may not be desirable or acceptable in another. Therefore, no single set o f criteria can define “a good requirements document”. The thesis presents a grounded theory which attempts to explain the diversity of styles of requirements documents found in practice, in relation to the variety of situations in which software products and systems are developed. It identifies the factors that might be useful to categorise situations from the point of view of requirements documentation. Requirements documents are widely used in software development, an activity typically carried out in an organisational context. Organisational theory suggests that the best approach in any situation depends on the factors that affect that situation. In the research, it was found that experienced practitioners employ a wide variety of constituent elements, structures, and styles when documenting requirements. This is in contrast with much of the literature on requirements engineering. The contribution o f this research is in three parts (a) an analysis o f requirements documents as texts, (b) a scheme for classifying system development situations with respect to the requirements documentation process, and (c) a framework matching typical requirements documents with the types o f situations identified in (a). As a grounded theory, it is the result of a detailed and systematic investigation into the role of requirements documents in the practice of software development Its status as a theory implies that it is tentative and provisional. An outline of how the theory might be validated for its usefulness, applicability, and generality is presented in the concluding chapter.