The purpose of this study was to investigate beginning teacher attrition and the factors of collaboration and school setting (urban, suburban, and rural). Quantitative data were collected from present and former teachers in a Midwest state using the researcher-created Collaboration Survey. Building principals and central office staff supplied attrition data from the 1998-2002 school years. Semi-structured follow up interviews with five percent of the sample population and unsolicited written comments found on the Collaboration Survey provided qualitative data. Aggregate computation of the four-year attrition rate revealed a substantial difference in the attrition rate by school setting. Urban schools had a much lower rate (4%) than did suburban (15%) and rural (17%) schools in the sample. A MANOVA was utilized to examine differences in the degree of collaboration experienced by current beginning teachers and former teachers with less than five years experience. Of the six subscales measuring collaboration, current and former teachers' total responses did not differ significantly. However, a one-way MANOVA was calculated for the three items comprising two subscales, Commitment and Process. Significant differences were found in one item from each of the two subscales. First, an item concerning the effect of the collaborative group on the beginning teacher's desire to continue teaching, and second, an item concerning the collaborative group member's responsibility for a common group of students. A Chi-Square test of independence determined that interdisciplinary teaming did not affect the attrition rate of beginning middle level teachers. (Contains 1 table.)