This paper describes an after-action review (AAR) behavior scoring system developed for a study of a medical simulation center's support to National Guard small team clinical training. The application of this AAR scoring system provides empirical insight into the behaviors key to AAR success. A significant part of modern training, the AAR is a professional, facilitated discussion within a training audience that compares trainee performance against task standards and training objectives. A great deal of study and theory has been developed on the conduct of an AAR. However, there are no quantifiable measures for enumerating the effectiveness of an AAR and AAR behaviors. Such measures would be valuable in determining the pertinent aspects of the AAR that facilitate the most effective experience for the training audience. The methodology presented in this paper, based on established AAR learning theory, describes fifteen AAR facilitator behaviors organized across four AAR elements as well as five training audience AAR behaviors and five overall AAR measures of performance. Two analysts scored thirty-two separate, small team AARs across these twenty-five measures. The scoring included both a simple count of behaviors and a behavior quality score. The scoring yielded an inter-rater agreement (Kappa) of 0.53, which is a moderate strength of agreement, instrumental in validating the scoring process. The paper covers the development, execution, and analytical results of the scoring method. The results indicate the specific behaviors most correlated to overall AAR success as well as contradicting some conventional wisdom regarding the AAR process. Implications of these findings on future AAR evaluation research are addressed.