APPROXIMATELY 12 MILLION Americans undergo spinal manipulation therapy (SMT) every year. Renewed interest in this method requires an analysis of its reported risks and possible benefits. This review describes two patients with spinal cord injuries associated with SMT and establishes the risk/benefit ratios for patients with lumbar or cervical pain. The first case is a man who underwent SMT for recurrent sciatica 4 years after chemonucleolysis. During therapy, he developed bilateral sciatica with urinary hesitancy. After self-referral, myelography demonstrated a total block; he underwent urgent discectomy with an excellent result 3 months after surgery. The second patient with an indwelling Broviac catheter and a history of lumbar osteomyelitis underwent SMT for neck pain. Therapy continued for 3 weeks despite the development of severe quadriparesis. After self-referral, he underwent an urgent anterior cervical decompression and removal of necrotic bone and an epidural abscess with partial neurological recovery. An analysis of these cases and 138 cases reported in the literature demonstrates six risk factors associated with complications of SMT. These include misdiagnosis, failure to recognize the onset or progression of neurological signs or symptoms, improper technique, SMT performed in the presence of a coagulation disorder or herniated nucleus pulposus, and manipulation of the cervical spine. Clinical trials of SMT have been summarized in several recent articles. Although these reviews agreed that most trials exhibited serious flaws, the data suggest that SMT demonstrates consistent effectiveness as an alternate treatment for adults with acute low back pain. SMT has not been shown to be superior to other conservative methods, nor to offer long-term benefits. It is concluded that the risk/benefit ratio is acceptably low for SMT as therapy for adults with midline low back pain of less than 1 week in duration. The ratio was unacceptably high for patients with radicular symptoms or signs associated with prolapsed discs and neck pain. Potential complications and unknown benefits indicate that SMT should not be used in the pediatric population.