Breast cancer is the most common cause of metastatic epidural spinal cord compression (SCC) in women, and this condition results in significant neurologic dysfunction and morbidity. Prior studies of patients with suspected SCC did not employ multivariate analysis techniques, often included persons with a wide variety of malignancies, and generally focused on identifying associated neurologic and radiologic features. We therefore conducted a study examining a more comprehensive set of potential clinical risk factors in breast cancer patients with suspected SCC. We retrospectively analysed 123 episodes of suspected SCC among 93 breast cancer patients evaluated by spine computed tomography (CT) scanning. Multiple logistic regression analysis was employed to identify independent predictors of SCC. Clinically significant metastatic epidural cancer was defined as thecal sac compression (TSC), which occurred in 33 episodes (27%). Four independent predictors of TSC were identified and included oncologic features (known bone metastases > or = 2 years, metastatic disease at initial diagnosis) in addition to neurologic and radiologic features (objective weakness, vertebral compression fracture on spine radiograph). These four predictors stratified episodes into subgroups with widely varying risks of TSC, ranging from 12% (0 risk factors) to 85% (> or = 3 risk factors). These results suggest that the evaluation of breast cancer patients with suspected SCC should include clinical information about their disease course in addition to neurologic examination and prior imaging studies. If confirmed, these predictors may help clinicians assess risk in this patient population.