As the expectations for student achievement continues to increase at the national, state, and local levels, the pressure specifically increases for classroom teachers to reflect and improve their instructional practices. The purpose of this study was to examine how peer coaching may promote and encourage metacognition and reflective practice, as a way to improve classroom instruction, and perhaps have a positive influence on student achievement (Bandura, 1994). The study identified elementary school teachers in Grades Pre-Kindergarten through Grade Six in public elementary schools in north Alabama that were participating in a peer coaching model called the Instructional Partnership Pilot (IPP) and assessed their levels of reflection (Gassenheimer, 2012, 2013). The teacher reflection levels were studied in two matched groups: one group of six elementary schools that had participated in a peer coaching model for either one, two, or three years and the other group was comprised of six elementary schools that had not participated in a formal peer coaching model. The study attempted to answer the following research questions: (1) Are there differences in the professional reflections of teachers involved in a peer coaching model as compared to the reflections of those teachers not involved in a peer coaching model? (2) Are there differences in the professional reflections of teachers involved in a peer coaching model such as the Instructional Partnership Pilot for one, two, or three years? The Reflective, Ethical, Moral Assessment Survey (REMAS, Arredondo Rucinski & Baugh, 2006) was used to assess the teacher perceptions. The data were then compared using a t-test and a multivariate analysis of variance (MANOVA). The results did not reveal a significant difference between the reflective perceptions of those that had and those that had not participated in peer coaching. However, it was determined that the teacher perceptions were statistically significantly different when comparing the number of years in the peer coaching model, more specifically with an implementation dip in year two and highest scores in year three.