Gamma-hydroxybutyric acid (GHB) is a naturally occurring transmitter in the mammalian brain, related to sleep regulation and possibly to energy balance in diving or hibernating animals. It has been used for almost 35 years as an intravenous agent for induction of anaesthesia and for long-term sedation. Its convincing pharmacological properties, without serious adverse effects on circulation or respiration, are compromised by its unpredictable duration of action. This is not a major problem with long-term sedation during ICU treatment. GHB has been used with good results for sedation of patients with severe brain injury, where it compares favourably with barbiturates. In animal studies, it seems to possess a protective action against hypoxia on a cellular and whole organ level. However, in some experimental animals GHB has been shown to produce seizure-like activities, and the compound is being used to produce absence-like seizures. GHB has been used in our ICU for years to provide adequate sedation for patients under controlled ventilation or for patients fighting the respirator during spontaneous respiration. No serious side effects were observed in these patients, while in some patients under haemodialysis hypernatraemia and metabolic alkalosis developed; both were reversible after discontinuation of GHB and restriction of additional sodium input (Somsanit, the commercially available GHB preparation in Germany, contains 9.2 mmol sodium/g; the daily dose averages 20-40 g GHB, i.e. 180-370 mmol sodium). PATIENTS AND METHODS. In 31 patients after major abdominal surgery, sedation was established with GHB 50 mg/kg BW injected via perfusion pump over a 20-min period. No centrally acting medication had been given for at least 2 h. A computer-based multichannel EEG system (CATEEM, MediSyst, Linden) was used, allowing for fast Fourier transformation, spectral analysis and topographical brain mapping. EEG during induction of sedation was followed after a baseline EEG (10 min) had been recorded. Patients receiving long-term sedation were studied daily for an additional 15-min period. Corresponding well to the clinical findings, EEG pattern changed to a slow delta-theta or delta-only rhythm within 10 min of the start of injection. Alpha and beta power decreased, while delta activity exhibited an increase. All changes were most obvious in frontal and central areas of the brain. In about one out of three patients, a burst--suppression pattern developed. Since automatic processing of EEG may fail to detect special patterns like the looked-for 3/s spikes and waves, the raw EEG was analysed visually by an expert neurologist. Both processed and conventionally analysed EEG were free of any seizure-like electrical activity. CONCLUSION. We conclude that animal data may not apply to the use of GHB in humans, provided the dose is limited to the clinical needs. GHB is used in clinical practice in doses twice as high, or even higher, than the one we use for induction, without obvious side effects. However, the suppression of theta rhythm we observed in about half of the patients studied may indicate that even less than 50 mg/kg BW might be sufficient for adequate sedation.