Despite substantial improvements in the last few years in software engineering and collaboration tools, coordination in large-scale software development continues to be problematic. This coordination is important because of the complex interdependencies that exist among software tasks, in that small productivity improvements can lead to substantial cost-savings and competitiveness. Traditional theories suggest that collaborators coordinate by organizing tasks and communicating, but recent research suggests that they also coordinate via implicit mechanisms like shared mental models. However, most of the shared mental model research literature focuses on real-time tasks, and there is very little empirical evidence on how these models affect coordination in more asynchronous and geographically distributed collaboration. Furthermore, none of this evidence is based on large-scale software development organizations. The present research is a field study at a large telecommunications company. It employs qualitative, quantitative, and survey research methods to investigate the effect of shared mental models on coordination in large-scale software development, and to better understand how geographic distance affects coordination.