An analysis of oculocephalic response and vestibuloocular reflex (VOR) in 81 patients with coma from various causes showed the importance of reflex eye movements for outcome prediction. Compared with oculocephalic response testing, VOR analysis provided more evidence and allowed more precise study of ocular motility in comatose patients. In 25 patients (31%) without conclusive or with absent oculocephalic response, a preserved VOR could be seen. The results indicated that in all cases with preserved VOR response (independent of the cause of coma) the assumption of a good outcome is justified and was correct in a maximum of 67%. The unpredictability in the other cases was largely due to non-neurological factors. Ninety two per cent of the patients with abolished reflex eye movements died. The combination of absent VOR and abolished pupillary light reaction allows prediction of negative outcome in 100% and shows the paramount importance of these two brainstem reflexes.