There has been sparse study of how requirements analysis is performed in situ, i.e. by organizations building large, complex systems. We assert that a better understanding of in situ requirements practice is necessary to improve and ground current theories of requirements analysis. We performed an empirical field study to examine in detail the issues faced by practitioners in forming and stabilizing requirements and the procedures they created to overcome them. We found that the process of requirements determination was intimately related to project selection. We further observed that these two processes were based on the interplay of 1) applied technical, project, and organizational authority with 2) design, sensemaking, and negotiation activity. The ethnography produced an idealized, grounded authority-activity model of requirements analysis and project selection. The model is a generalized form of requirements analysis-project selection for large, complex, risk adverse (highly sensitive to failure) projects. It represents a method to balance the differentiated authority of the stakeholder groups with the activities necessary to form and stabilize technology-project candidates and their related requirements. We argue that the core issues addressed in this field study are generalized across organizations that build large, complex systems and hence, the results of this study form a basis for a general theory of requirements analysis practice.