We conducted a prospective cohort study to determine whether administration of large doses of the corticosteroid methylprednisolone following spinal cord injury as recommended in the National Acute Spinal Cord Injury Study-2 (NASCIS-2) protocol results in an increased incidence of avascular necrosis (AVN) of the femoral or humeral head. All subjects were patients treated by a spinal cord injury physician in an Acute Spinal Cord Injury Unit between 1989 and 1996 where some received the megadose steroids while others did not. Patients younger than 15 years and older than 75 years were excluded, as were those with any hip or shoulder disease, with pelvic fracture, or with a history of predisposition to AVN by hip dislocation, excessive alcohol consumption, previous high dose steroid use, or systemic lupus erythematosus. Screening for AVN of the femoral and humeral heads was performed at a minimum of 6 months following injury, using magnetic resonance imaging (MRI). The films were read by a radiologist blinded to the treatment protocol received by the individual subject. Among the 59 spinal cord injured patients who received steroids (age 15-64 years (mean 32 years)), five were female. Among the 32 spinal cord injured subjects who did not receive steroids (age 16 to 65 years (mean 34 years)), seven were female. There was no case of AVN found in either group. Using binomial distribution, we conclude that the true incidence of AVN among the methylprednisolone treated group is less than 5% (alpha < 0.05) and therefore continue to recommend short term (24 h) methylprednisolone therapy.