In an effort to determine the information needs of tower air traffic controllers, instructors from the Federal Aviation Administration's Academy in Oklahoma City were asked to control traffic in a high-fidelity tower cab simulator. Information requests were made apparent by eliminating access to standard tower information sources. Instead, controllers were required to ask for precisely the information they needed during the scenarios. The information requests were classified using an elaboration of Zwaan and Radvansky's (1998) dimensions of situation models. The vast majority of requests were about three of the dimensions originally developed for reading comprehension: the protagonist, intentionality, and space. The information requests were also classified into 28 operational categories (e.g., aircraft identification, destination). From these results, the data were summarized, not just statistically, but by the creation of display-hypotheses. The display-hypotheses were organized according to the situation-model dimensions. Controllers preferred data blocks organized by the situation-model principle over those that violated this organization. The summary display-hypotheses were quite simple and accounted for the vast majority of the information requests controllers made. The display-hypotheses accounted for the information needs of controllers during routine as well as off-nominal events. (Contains 6 tables and 6 figures.)