The aims of this study were to document the efficacy of treatment and to identify factors that were predictive of the outcome in malignant epidural spinal; cord compression. The medical records of patients treated at the Prince Henry and Prince of Wales Hospitals in the period 1980-1989 with a diagnosis of malignant epidural spinal cord compression were reviewed. A total of 94 patients were eligible for the study and were treated by radiotherapy alone (37), surgery alone (19) and surgery followed by radiotherapy (38). Efficacy was determined by measuring complete resolution of symptoms and signs at 1 month after presentation, and also by using an overall functional improvement score (FIS). Complete resolution of individual pre-treatment symptoms that were measured 1 month after treatment occurred as follows: pain (30/88), sensory disturbance (12/61), weakness (8/17), bladder dysfunction (10/42), and bowel dysfunction (10/36). Complete resolution of motor deficit occurred in 7/82 and of sensory deficit in 9/73. The ability to walk was regained in 19/51 previously non-ambulatory patients, and bladder function improved sufficiently to remove an indwelling catheter in 9/32 previously catheterized patients. As judged by FIS, 67 patients improved, 15 patients remained stable and 12 patients deteriorated. Of the treatments given, a combination of surgery followed by radiotherapy was associated with the greatest functional improvement (P = 0.001). The coexistence of 'liver failure' was the only patient-related factor identified which was associated with outcome (P = 0.041). The treatment of malignant spinal cord compression appears to be worthwhile; however, the outcome of treatment is not easy to predict from pretreatment factors. A 'functional improvement score' may be useful in assessing treatment efficacy.