METHODS : Prediction study. OBJECTIVE : To investigate the prediction accuracy of the Downton fall risk index (DFRI) and a question of falls the previous year; further to examine the association between time to first fall and risk indicators for falls in wheelchair users and ambulatory persons with Spinal Cord Injury (SCI). METHODS : Two SCI centres in Norway and Sweden (Sunnaas Rehabilitation Hospital, Rehab Station Stockholm /Spinalis). METHODS : Two hundred and twenty-four persons with traumatic SCI, ≥1 year post-injury, ≥18 years participated. Prospective falls were reported by text messages every second week for one year. Sensitivity, specificity and time to first fall (Kaplan Meier) were investigated for DFRI and the question of falls in the previous year. DFRI ≥3 was defined as a high risk of falls. Cox survival analysis was used to calculate hazard ratios for functional independence, gait speed and fear of falling. RESULTS : The sensitivity was 36-57% for DFRI and 82-89% for the question of falls, while specificity was 74-83%, and 34-49%. For DFRI, time to first fall was shorter in the high-risk group, for both wheelchair users (p = 0.005) and ambulatory persons (p = 0.006). Falls previous year increased the hazard ratio of falls for wheelchair users (HR = 3.35, 95% CI = 1.86 to 6.02) but not for ambulatory persons. CONCLUSIONS : Falls in the previous year showed a better predictive accuracy than DFRI due to the low sensitivity of DFRI. As previous falls had low specificity, it still remains difficult to predict falls in those who have not fallen yet.