Cortical spreading depolarizations occur spontaneously after ischaemic, haemorrhagic and traumatic brain injury. Their effects vary spatially and temporally as graded phenomena, from infarction to complete recovery, and are reflected in the duration of depolarization measured by the negative direct current shift of electrocorticographic recordings. In the focal ischaemic penumbra, peri-infarct depolarizations have prolonged direct current shifts and cause progressive recruitment of the penumbra into the core infarct. In traumatic brain injury, the effects of spreading depolarizations are unknown, although prolonged events have not been observed in animal models. To determine whether detrimental penumbral-type depolarizations occur in human brain trauma, we analysed electrocorticographic recordings obtained by subdural electrode-strip monitoring during intensive care. Of 53 patients studied, 10 exhibited spreading depolarizations in an electrophysiologic penumbra (i.e. isoelectric cortex with no spontaneous activity). All 10 patients (100%) with isoelectric spreading depolarizations had poor outcomes, defined as death, vegetative state, or severe disability at 6 months. In contrast, poor outcomes were observed in 60% of patients (12/20) who had spreading depolarizations with depression of spontaneous activity and only 26% of patients (6/23) who had no depolarizations (χ2, P<0.001). Spontaneous electrocorticographic activity and direct current shifts of depolarizations were further examined in nine patients. Direct current shift durations (n=295) were distributed with a significant positive skew (range 0:51-16:19 min:s), evidencing a normally distributed group of short events and a sub-group of prolonged events. Prolonged direct current shifts were more commonly associated with isoelectric depolarizations (median 2 min 36 s), whereas shorter depolarizations occurred with depression of spontaneous activity (median 2 min 10 s; P<0.001). In the latter group, direct current shift durations correlated with electrocorticographic depression periods, and were longer when preceded by periodic epileptiform discharges than by continuous delta (0.5-4.0 Hz) or higher frequency activity. Prolonged direct current shifts (>3 min) also occurred mainly within temporal clusters of events. Our results show for the first time that spreading depolarizations are associated with worse clinical outcome after traumatic brain injury. Furthermore, based on animal models of brain injury, the prolonged durations of depolarizations raise the possibility that these events may contribute to maturation of cortical lesions. Prolonged depolarizations, measured by negative direct current shifts, were associated with (i) isoelectricity or periodic epileptiform discharges; (ii) prolonged depression of spontaneous activity and (iii) occurrence in temporal clusters. Depolarizations with these characteristics are likely to reflect a worse prognosis.