Chest pain may be caused by joint and muscle dysfunction of the neck and thorax (termed musculoskeletal chest pain). The objectives of this study were (1) to determine inter-observer reliability of the diagnosis 'musculoskeletal chest pain' in patients with acute chest pain of non-cardiac origin using a standardized examination protocol, (2) to determine inter-observer reliability of single components of the protocol, and (3) to determine the effect of observer experience. Eighty patients were recruited from an emergency cardiology department. Patients were eligible if an obvious cardiac or non-cardiac diagnosis could not be established at the cardiology department. Four observers (two chiropractors and two chiropractic students) performed general health and manual examination of the spine and chest wall. Percentage agreement, Cohen's Kappa and ICC were calculated for observer pairs (chiropractors and students) and all. Musculoskeletal chest pain was diagnosed in 45 percent of patients. Inter-observer kappa values were substantial for the chiropractors and overall (0.73 and 0.62, respectively), and moderate for the students (0.47). For single items of the protocol, the overall kappa ranged from 0.01 to 0.59. Provided adequate training of observers, the examination protocol can be used in carefully selected patients in clinical settings and should be included in pre- and post-graduate clinical training.