The authors report the clinical symptoms and response to therapy of a series of patients who presented with subacute or chronic back pain due to vertebral osteonecrosis (Kummell's spondylitis) and who underwent percutaneous vertebroplasty. The authors performed a retrospective chart review of a series of 95 patients in whom 149 painful, nonneoplastic compression fractures were demonstrated and who were treated with percutaneous transpediculate polymethylmethacrylate (PMMA) vertebroplasty. In six of these patients there was evidence of vertebral osteonecrosis, as evidenced by the presence of an intravertebral vacuum cleft on radiography or by intravertebral fluid on magnetic resonance (MR) imaging. Clinical and radiological findings on presentation were noted. Technical aspects of the vertebroplasty technique were compiled. Response to therapy, defined as qualitative change in pain severity and change in level of activity, was noted immediately following the procedure and at various periods on follow-up reviews. One man and five women, who ranged in age from 72 to 90 years (mean 81 years), were treated. Each patient had one compression fracture. The fractures were at T-11 (one patient), L-1 (two patients), L-3 (two patients), and L-4 (one patient). The pain pattern was described as severe and localized to the affected vertebra, and sometimes radiated along either flank. Pain duration ranged from 2 to 12 weeks, and the pain was refractory to conservative therapy that consisted of bedrest, analgesics, and external bracing. At the time of treatment, all patients were bedridden because of severe back pain. In all patients either plain radiographic or computerized tomography evidence of intravertebral vacuum cleft or MR imaging evidence of vertebral fluid collection consistent with avascular necrosis of the vertebral body was demonstrated. Four patients underwent bilateral transpediculate vertebroplasty, and two patients underwent unilateral transpediculate vertebroplasty. The fracture cavities were specifically targeted for PMMA injection. Additional fortification of the osteoporotic vertebral body trabeculae was also performed when feasible. "Cavitygrams" or intraosseous venograms with gentle contrast injection were obtained prior to application of cement mixture. In all patients subjective improvement in pain and increased mobility were demonstrated posttreatment. The follow-up period ranged from 4 to 24 hours after treatment. Two patients made additional office visits at 1 and 3 months, respectively. Patients presenting with vertebral osteonecrosis (Kummell's spondylitis) often suffer from local paraspinous or referred pain. When performing vertebroplasty on these patients, confirmation of entry into the fracture cavities with contrast-enhanced "cavitygrams" should be performed prior to injection of PMMA cement. The response to vertebroplasty with regard to amelioration of pain and improved mobility is encouraging.