High teacher turnover in large U.S. cities is a critical issue for schools and districts, and the students they serve; but surprisingly little work has been done to develop methodologies and standards that districts and schools can use to make reliable estimates of turnover costs. Even less is known about how to detect variations in turnover costs for teachers of different grades and disciplines. This study created a model and methodology to document turnover costs for the middle and high schools in the Boston Public Schools to test the degree to which it could detect differences in costs for teachers of science, and to explore the feasibility of its implementation by school personnel. We found that although the model and methods were sensitive enough to detect differences between schools and for science teachers, it could not be easily or fully applied at the district or school levels, where the component costs of turnover were scattered between department budgets, and where some costs were undocumented. At the same time, our data show that when teachers leave a school, whether it is due to a departure from the district or a reassignment to another school, there is an associated cost that has previously gone unmeasured. Therefore, understanding the full cost of teacher departures is essential when trying to estimate the true cost--and cost savings--of a layoff. Since professional development consistently represents the largest cost in the district studied, we suggest steps that can be taken to track these expenses more systematically at the school and district levels. (Contains 5 tables, 1 figure and 13 footnotes.)