Decision-making in large, information-intensive domains is discussed in terms of information gathering strategies employed by experienced, intermediate, and novice decision-makers. Differences by level of experience were predicted for seven strategies. These predictions were tested in a process-tracing experiment using a submarine search scenario. Differences were found among groups in the use of information gathering strategies, in response to written questions, and in the content of recorded verbal protocols. The experimental results were used to build an expert decision aid that adapts dynamically to the experience level of its user. This is done by tracking information-gathering behavior and adapting the system response when the behavior is not expert-like. An unusual aspect of this system was the use of the situational model, to model not only the situation but also user behavior. >