UNLABELLED Until recently, the U.S. Army Combat Medic School used a traditional teaching model with heavy emphasis on large group lectures. Skills were taught separately with minimal links to didactics. OBJECTIVES To evaluate whether the adult learning model improves student learning in terms of cognitive performance and perception of proficiency in military medic training. METHODS The study population was two sequential groups of randomly selected junior, enlisted, active duty soldiers with no prior formal emergency medical training who were enrolled in an experimental model of a U.S. Army Combat Medic School. The control population was a similar group of students enrolled in the traditional curriculum. Instructors were drawn from the same pool, with experimental group instructors receiving two weeks of training in adult-learning strategies. The study population was enrolled in the experimental program that emphasized the principles of adult learning, including small-group interactive approach, self-directed study, multimedia didactics, and intensive integrated practice of psychomotor skills. Instructors and students were also surveyed at the end of the course as to their confidence in performing four critical skills. The survey instrument used a five-point scale ranging from "strongly disagree" through "undecided" to "strongly agree." Proficiency for this survey was defined as the sum of the top two ratings of "agree" or "strongly agree" to questions regarding the particular skill. Both experimental and control programs lasted ten weeks and covered the same academic content and nonacademic (e.g., physical fitness) requirements, and the two groups of students had similar duty days. Evaluations included performance on internal and National Registry of Emergency Medical Technicians (NREMT) written examinations and other measures of academic and nonacademic performance. RESULTS One hundred fifty students (experimental n = 81, control n = 69) were enrolled in 1999-2000. The scores for internal course grade, NREMT written score, and NREMT written pass rate were, respectively, 86.3, 71.6, and 63% for the experimental group; and 85.8, 69.6, and 49% for the control group. The p-value was </= 0.05 for the comparison between internal course grade and NREMT written score, but p > 0.05 for the comparison between NREMT written pass rates. Students in both the adult-learning and traditional groups rated themselves high in proficiency, whereas instructors in the traditional group were generally much more modest in their rating of student proficiency than instructors of the adult-learning program. CONCLUSIONS In this study setting, an adult-learning model offers only a modest improvement in cognitive evaluation scores over traditional teaching when measured at the end of the course. Additionally, students in the traditional teaching model assess themselves as proficient more frequently than instructors, whereas instructor and student perception of proficiency more closely matched in the adult-learning model.