Electrocorticographic (ECoG) activity was recorded for up to 129 h from 12 acutely brain-injured human patients using six platinum electrodes placed near foci of damaged cortical tissue. The method probes ECoG activity in the immediate vicinity of the injured cortex and in adjacent supposedly healthy tissue. Six out of twelve patients displayed a total of 73 spontaneous episodes of spreading depression of the ECoG. Of the remaining 6 patients 1 displayed an episode of synchronous depression of ECoG during surgery. Using the same electrodes we also measured the slow potential changes (SPC) (0.005-0.05 Hz) to test the hypothesis that the ECoG depressions were identical to Leao's cortical spreading depression (CSD), and to be able to record peri-infarct depolarisations (PIDs) in electrically 'silent' cortical tissue. Changes in the SPC indicate depolarization of brain tissue. For the analysis, the SPCs were enhanced by calculating the time integral of the ECoG signal. Spreading ECoG depressions were accompanied at every single recording site by stereotyped SPCs, which spread across the cortical mantle at 3.3 (0.41-10) mm/min (median, range), i.e. at the same speed of spread as the depression of the ECoG activity. The amplitude of the SPCs was 0.06-3 mV. In 4 out of 6 patients the ECoG recovered spontaneously. In 2 patients we subsequently recorded recurrent SPCs, but without recovery of the initial ECoG background activity until 2-5 h later. This represents the first direct recording of PIDs in acutely injured human brain. Evidence from this and our previous study of 14 brain-injured patients suggests that CSDs in acute brain disorders occur at higher incidence in patients <30 years (83%) than above (33%). CSD was recorded in 4 out of 5 traumatic brain injury patients, and in 2 out of 7 patients with spontaneous haemorrhages. We conclude that the spreading ECoG depressions recorded in patients are identical to CSDs recorded in animal experiments. We furthermore provide direct electrophysiological evidence for the existence of PIDs and hence a penumbra in the human brain. We hypothesize that the depolarization events might contribute to tissue damage in acute disorders in the human brain.