Study design: Systematic literature review for primary data using prede®ned inclusion, exclusion and validity criteria. Primary outcome measure was standardised neurological examination or neurological function. Secondary outcomes; acute mortality, early morbidity. Objectives: To access the literature available to clinicians systematically and evaluate the evidence for an eect of high dose methylprednisolone (MPSS) on neurological improvement following acute spinal cord injury (ACSI). Methods: Information retrieval was based on Medline search (1966 through December 1999) using the strategy`spinal cord injury' and`methylprednisolone' (or`dexamethasone') with no other restrictions. Primary data publications using high dose steroids given within 12 h following spinal cord injury and reporting outcome measures separately for steroid and nonsteroid treated groups were selected. Evaluation followed the guides of Guyatt et al 7 (for the Evidence Based Working Group in Canada). Studies with questionable validity were excluded. Level of evidence and treatment recommendation utilised the Canadian Task Force on the Periodic Health Examination criteria. 6 Experimental spinal cord injury studies on larger animals were included; small mammal experiments were considered beyond evaluation. Results: Three clinical trials and six cohort study publications were found to satisfy the review criteria. The evidence they provide supports`the recommendation that the manoeuvre (high dose methylpredisolone) be excluded from consideration as an intervention for the condition' 10 (acute spinal cord injury). Twelve larger animal publications were detailed. Validity and the functional signi®cance of results was of concern in many. The weight of evidence lay with those studies demonstrating no de®nite eect of MPSS on functional outcome. In cat experiments with higher level cord damage, deaths in the MPSS treated groups were notable. Conclusion: The evidence produced by this systematic review does not support the use of high dose methylprednisolone in acute spinal cord injury to improve neurological recovery. A deleterious eect on early mortality and morbidity cannot be excluded by this evidence. Spinal Cord (2000) 38, 273 ± 286