Study design: Mortality review was undertaken of patients who su ered traumatic spinal cord injury (SCI) between 1955 and 1994 inclusive. Objectives: The study objective was to provide evidence of reasons for the observed reduction in long-term life expectancy for the SCI population. Setting: Patients were those who had most, if not all, of their inpatient and outpatient care at Royal North Shore Hospital, Spinal Injuries Unit, Sydney, New South Wales, Australia. Methods: Data on causes of death for 195 patients ®tting the inclusion criteria were analysed by actuarial methods using ICD9CM classi®cations. Results: The incidence of death in the spinal cord injured, from septicaemia, pneumonia and in¯uenza, diseases of the urinary uystem and suicide, are signi®cantly higher than in the general population. The ®ndings con®rm variations in potentially treatable causes of death depending on neurological impairment, attained age and duration since injury. Unlike septicaemia and pneumonia, which have shown a signi®cant reduction since 1980, the death rate for suicide alone has risen. Conclusion: This analysis identi®ed complications which a ect mortality and morbidity in patients su ering from the e ects of SCI. Spinal Cord (2000) 38, 604 ± 610