Patients with incomplete spinal cord injuries can spontaneously recover motor function. Because of this, phase I and II trials of invasive interventions for acute spinal cord injury will likely involve neurologically complete injuries. It is therefore important to reliably identify complete injuries as early as possible. We examined the reliability of the early examination in motor complete spinal cord injuries by retrospectively analyzing the stability of baseline neurological status determined within 2 days of injury in 103 subjects. Baseline neurological status was compared to neurological status at follow-up, preferably within one week (101 of 103 subjects). When available (n = 68), neurological status at 1 year or later was also compared. Overall, 6.2% (5/81) of motor complete, sensory complete (ASIA A) subjects converted to motor complete, sensory incomplete status (ASIA B) between the initial and follow-up assessments; however, none exhibited motor recovery (ASIA C or D). At initial follow-up, 9.3% (4/43) of ASIA A subjects with factors affecting examination reliability were reclassified as ASIA B injuries compared to 2.6% (1/38) of ASIA A subjects without such factors. At year 1 or later, 6.7% (2/30) of ASIA A subjects without factors affecting exam reliability, converted to ASIA B status. None developed volitional motor function below the zone of injury. For subjects with factors affecting exam reliability, 17.4% (4/23) of ASIA A subjects converted to incomplete status and 13.0% (3/23) regained some motor function by one year or later (ASIA C or D). These data suggest that it is possible to identify within 48 h of injury, a subset of patients with a negligible chance for motor recovery who would be suitable candidates for future clinical trials of invasive treatments.