Fifty-six consecutive patients clinically diagnosed as brain dead were seen at Loyola University Medical Center, Maywood, Ill, from January 1984 through May 1986. Eleven (19.6%) of the 56 patients had electroencephalographic (EEG) activity following the diagnosis of brain death. The mean duration of the observed EEG activity was 36.6 hours (range, two to 168 hours). Three patterns of EEG activity were observed: (1) low-voltage (4 to 20 microV) theta or beta activity was recorded in nine (16.1%) patients as long as 72 hours following brain death. Neuropathologic studies in one patient showed hypoxic-ischemic neuronal changes involving all cell layers of the cerebral cortex, basal ganglia, brain stem, and cerebellum; (2) sleep-like activity (a mixture of synchronous 30 to 40 microV theta and delta activity and 60 to 80 microV, 10 to 12 Hz spindle-like potentials) was noted in two (3.6%) patients for as long as 168 hours following brain death. Pathologic studies in both cases demonstrated ischemic necrosis of the brain stem with relative preservation of the cerebral cortex; and (3) alpha-like activity (monotonous, unreactive, anteriorly predominant, 25 to 40 microV, 9 to 12 Hz activity) was observed in one (1.8%) patient three hours following brain death. Regardless of activity on the EEG, none of the patients recovered. The occurrence of EEG activity following brain death would suggest reliance on the EEG to confirm brain death may be unwarranted. The use of the EEG as a confirmatory test of brain death may be of questionable value.