STUDY DESIGN: In vitro biomechanical study of human cadaveric thoracic spine segments and one intact cadaver and applied load measurements in human volunteers. OBJECTIVES: To quantify failure load and pattern of midthoracic vertebrae under a posteroanterior load and to compare failure load in vitro with applied load in vivo. SUMMARY OF BACKGROUND DATA: Osteoporosis and back pain are common alone and in combination among older adults. Spinal mobilization techniques have been shown to relieve back pain and improve function in various clinical settings. However, whether controlled spinal mobilization can cause vertebral fracture in individuals with osteoporosis is not known. METHODS: Twelve T5-T8 cadaveric specimens (mean age, 77 years) were scanned using bone densitometry, radiographed, and measured for bone size. The authors measured failure load, failure site, and intervertebral motion (using a precision optoelectronic camera system) when a posteroanterior load was applied at the spinous process of T6 using a servohydraulic material testing machine. Post-test radiography and CT scan were used to verify failure site. These tests were repeated in an intact cadaver using a Tekscan I-Scan sensor to measure applied loads. The authors also quantified in vivo applied loads during posteroanterior mobilization during seven trials by two experienced physiotherapists. RESULTS: Mean (SD) in vitro failure load of 479 N (162 N) was significantly higher than the mean (SD) in vivo applied load of 145 N (38 N) (P = 0.0004). Macroscopic observation revealed a fracture at the T6 spinous process in 11 specimens and one at the T7 spinous process. These fractures were detected by plain radiography in three of 12 cases and by CT scan in six of 12 cases. CONCLUSIONS: The results suggest a reasonable margin between failure load in vitro and applied mobilization load in vivo.