Study objective: The objectives of this study are to elicit and document descriptions of emergency physician expertise, to characterize cognitive differences between novice and expert physicians, and to identify areas in which novices' skill and knowledge gaps are most pronounced. The nature of the differences between novices and experts needs to be explored to develop effective instructional modalities that accelerate the learning curve of inexperienced physicians who work in high-complexity environments. Methods: We interviewed novice emergency physicians (first-year residents) and attending physicians with significant expertise, working in an academic Level I trauma center in Southern California. With cognitive task analysis, we used task diagrams to capture nonroutine critical incidents that required the use of complex cognitive skills. Timelines were constructed to develop a detailed understanding of challenging incidents and the decisions involved as the incident unfolded, followed by progressive deepening to tease out situation-specific cues, knowledge, and information that experts and novices used. A thematic analysis of the interview transcripts was conducted to identify key categories. Using classification techniques for data reduction, we identified a smaller set of key themes, which composed the core findings of the study. Results: Five interns and 6 attending physicians participated in the interviews. Novice physicians reported having difficulties representing the patient's story to attending physicians and other health care providers. Overrelying on objective data, novice physicians use linear thinking to move to diagnosis quickly and are likely to discount and explain away data that do not "fit" the frame. Experienced physicians draw on expertise to recognize cues and patterns while leaving room for altering or even changing their initial diagnosis. Whereas experts maintain high levels of spatial, temporal, and organizational systems awareness when overseeing treatment modalities of multiple patients, novices have difficulty seeing and maintaining the "big picture." Conclusion: Novice physicians use sense-making styles that differ from those of experts. Training novices to respond to the high cognitive demand of complex environments early in their careers requires instructional modalities that not only increase their knowledge base but also accelerate the integration of knowledge and practice. Simulation and custom-designed avatar-mediated virtual worlds are a promising new technology that may facilitate such training. Future research should expand on the results of this study through the use of larger sample sizes and interviews conducted at multiple sites to increase generalizability.