The ability to predict the resolution of post-traumatic amnesia (PTA) early in the course of acute rehabilitation was studied in a sample of 98 closed-head injury (CHI) patients admitted for acute rehabilitation. The subjects' age, length of coma, time post-injury and Orientation Group Monitoring System (OGMS) subscale and aggregate scores were evaluated for the ability to predict if and when a patient would clear PTA. The results indicated that the combination of time post-injury and week one OGMS aggregate score provided the best prediction of whether and when PTA would clear. A shorter time from injury to acute rehabilitation was a positive prognostic indicator, accounting for more variance than either age or duration of coma. Whereas the aggregate OGMS score was a better indicator than any of the subscale scores, the results also suggested that specific aspects of cognition have differential predictive power, which may derive from different types of injury to the brain. Although the prediction model was statistically significant, its practical significance was limited. In particular, the high base rate for clearing PTA, 83.7%, attenuated the potential for differential prediction. Finally, this study provided additional support for the construct validity of the OGMS as a measure of cognitive function during the acute phase of recovery from CHI.