Peer assisted learning (PAL) occurs in small groups and in dyads. It is most commonly seen in forms such as tutoring and cooperative learning, but it is also a component of instructional models such as Teaching Games for Understanding, Sport Education, and Social Responsibility. In addition, PAL has been recommended as a strategy to promote inclusion, equity, social skill development, cognitive learning and psychomotor learning (Barrett, 2000; Cohen, 1994; Dyson 2001; Siedentop & Tannehill, 2000). The research base for PAL is extensive, demonstrating good utility in such areas as general education, special education, higher education, and in non-academic areas such as after school programming (Cohen, 1994; De Lisi & Goldbeck; 1999, Slavin, 1990). Despite this utility, PAL is not a magic bullet. Merely placing students in groups is insufficient to ensure that learning will occur. This presentation will report on the evidence in physical education and general education literatures that suggests that PAL effects are not incidental, but rather part of explicit elements which are functionally related to the relationship between group and individual accountability; and where an essential instructional skill is the selection of tasks that create reciprocal interdependence among students. Examination of the literature in physical education reveals that PAL works differentially relative to the type of: (a) strategy, (b) task, (b) student grouping, (c) grade level, (d) gender, (e) the ability of the student , and (f) the level of training the students received to perform the strategy. The utility of these findings for teaching in physical education are discussed. The presentation concludes with suggestions for future directions in research.Keyword(s): curriculum development, professional development, research