Factors affecting survival were determined for 109 patients with thoracic spine metastases and cord compression. Lung, prostate, and breast were the most common primary sites (78%). All patients had surgical decompression of the spinal cord, and 99% received radiotherapy. Survival was determined based on anatomic site of primary carcinoma, preoperative neurologic deficit, extent of disease, number of vertebral bodies involved, tumor location (site of cord compression), and age. The respective compounding weight of the negative prognostic factors also was analyzed in terms of survival. The overall median survival was 10 months. Patients preoperatively ambulatory survived statistically significantly longer than nonambulatory patients or those with sphincter incontinence (P = 3.469 x 10(-6)). Patients with renal cell carcinoma survived the longest, followed by those with breast, prostate, lung, and colon cancer. Patients with breast cancer survived statistically longer than those with lung cancer (P = 0.039). Patients with one vertebral body involved survived statistically significantly longer than patients with multiple vertebral level involvement (P = 0.027). Extent of disease, age, and tumor location did not significantly influence survival. In patients with vertebral column disease, the presence of two or more poor prognostic indicators (leg strength 0/5-3/5, lung or colon cancer, multiple vertebral body involvement), had a compounding adverse effect on survival. For patients with spinal metastases and cord compression, the factors found to affect survival include preoperative neurological status, anatomic site of primary carcinoma, and number of vertebral bodies involved. Patients with vertebral column disease and two or more of the poor prognostic indicators have a short life expectancy, and, therefore, radical surgery is not recommended because the benefits may not be substantial.