Physical measures of the NIOSH Low Back Atlas (LBA) were used to distinguish groups of musculoskeletal low back pain (LBP) patients and asymptomatic controls. Subjects were 115 patients with LBP, and 112 referents of similar age and gender. The Vermont Low Back Pain Questionnaire, the SCL-90-R, the McGill Pain Questionnaire, the Roland Morris Scale and the Activities of Daily Living Scale were administered. The 25 items of the NIOSH LBA and three Waddell tests were given by ten physical therapists. Results showed that patients were distinguished into four cluster groups by measures of symmetry, flexibility, strength, and dynamic mobility. Referents fell into five cluster groups. Measures that were similar in the cluster groups were past history, smoking, handedness, nature of pain, duration of pain, intensity of pain, symptom magnification, and outcome. Those that were different were the Davenport index, age, gender, function, psychologic measures, and test refusals. The patient subgroup that had the most physical involvement also showed the most functional limitations and the most psychosocial involvement. Poorer self images and higher hostility levels probably did not contribute to the scores as each NIOSH test item could be refused. The authors conclude that the NIOSH LBA enables groups of LBP patients as well as asymptomatic controls to be distinguished.