Demand for acceptance-based performance enhancement in the military has increased. This demand has created a need for effective measurement of mental skills training interventions. There is evidence for measuring performance enhancement interventions with the Acceptance and Action Questionnaire-II (AAQ-II), a measure of psychological inflexibility (PI). The existence of multiple versions of the AAQ-II has caused confusion, and practitioners and researchers currently use both the 7- and 10-item versions. To address this need, responses from Special Operations Forces personnel were used to assess the psychometric properties of two versions of the AAQ-II, and to make a recommendation for the preferred model of the AAQ-II. In Study 1 (n = 416), the factor structure was examined using confirmatory factor analysis and alternative model testing. The AAQ-II was determined to be a unidimensional structure and analyses supported removal of the three reverse-scored items from the 10-item version of the scale. Removal of these items decreases cognitive load for participants and improves direct measurement of PI. Two problematic items were also removed, resulting in a 5-item model of PI. The recommended 5-item model demonstrated convergent and criterion validity, similar in magnitude and direction to the 10- and 7-item versions of the model. In Study 2 (n = 284), the retained 5-item model of PI fit a more diverse sample of Special Operations Forces personnel. The 5-item model demonstrated good internal reliability and demonstrated significant moderate relationships with convergent mindfulness factors. Future research should focus on creating response options which are more sensitive to differences in low levels of PI.