INTRODUCTION THE MAGNITUDE of the clinical problem of neoplastic spinal cord compression (SCC) is usually underestimated. The estimated annual incidence of cancer-induced spinal injury in the U.S.A. is 8.5 per 100 000 [ 11, which exceeds the calculated incidence of the annual rate of traumatic spinal cord injury (three to five per 100 000). Since SCC in itself is generally not fatal (excluding the upper portion of the cervical spine), treatment is aimed at preserving or restoring ambulation and continence, and alleviating intractable pain. Approximately 50% of the cases of metastatic epidural compression in adults arise from breast, lung or prostatic cancer [2-51, while in children the most common tumour types are different. Sarcomas and neuroblastomas comprise more than 80% of all cases of paediatric SCC [2, 6, 71, and most paediatric tumours invade the spinal canal via the neural foramen rather than via vertebral destruction which is the common mode for neoplastic epidural invasion in adults. These variances usually dictate therapeutic approaches tailored to match the different types of tumour, but the functional outcome still depends on the same pathophysiological mechanism operating in both age groups. The last two decades have witnessed significant shifts in therapeutic approaches, ranging between urgent decompressive surgery and non-surgical treatments [8]. Regardless of the treatment modality in use, less than 50% of all patients with SCC ever walk again [8], and even prompt decompression does not guarantee neurological recovery. The mechanism that determines the degree of irreversible tissue damage is poorly understood, but appears to be associated with endogenous neurochemical changes resulting from the initial event of compressive injury. Therefore, it seems crucial to link strategies that alleviate compressive mechanical injury with tactics designed to limit the secondary autodestructive processes operating in the compressed spinal cord and leading to neuronal cell death and permanent loss of function. Comprehension of the neurochemical cascade ‘activated by neoplastic compression has been expanded by the use of newly characterised animal models that enable better understanding of tissue injury at a cellular level [+21]. The recognition that