The authors present data describing the temporal progress of 786 severely head-injured patients from discharge to 12 months postinjury. Changes in outcome over time are characterized and the effects on these changes of patient age, Glasgow Coma Scale score, pupillary response, and type of injury are investigated. The feasibility of using these factors and the outcome at 3 months postinjury to predict the outcome at 6 months post-trauma is explored via logistic regression analysis. The data indicate that a significant proportion of patients improve continuously during the first 6 months after injury; outcome tends to stabilize thereafter, which suggests that the response at this time may be the appropriate end point for clinical trials in severe head injury. Dichotomized outcomes are predictable with approximately 94% accuracy at 6 months postinjury; therefore, missing outcomes may safely be replaced by the corresponding predicted outcomes. The findings also indicate striking improvement over time in patients who are in a vegetative state at discharge, suggesting that decisions regarding withdrawal of supportive care should be postponed until 6 months after injury in these patients.