Background. We have investigated the effect of propofol on language processing using event-related functional magnetic resonance imaging (MRI). Methods. Twelve healthy male volunteers underwent MRI scanning at a magnetic field strength of 3 Tesla while performing an auditory language processing task. Functional images were acquired from the perisylvian cortical regions that are associated with auditory and language processing. The experiment consisted of three blocks: awake state (block 1), induction of anaesthesia with 3 mg kg(-1) propofol (block 2), and maintenance of anaesthesia with 3 mg kg(-1) h(-1) propofol (block 3). During each block normal sentences and pseudo-word sentences were presented in random order. The subjects were instructed to press a button to indicate whether a sentence was made up of pseudo-words or not. All subjects stopped responding during block two. The data collected before and after the subjects stopped responding during this block were analyzed separately. In addition, propofol plasma concentrations were measured and the effect-site concentrations of propofol were calculated. Results. During wakefulness, language processing induced brain activation in a widely distributed temporofrontal network. Immediately after unresponsiveness, activation disappeared in frontal areas but persisted in both temporal lobes (block 2 second half, propofol effect-site concentration: 1.51 mug ml(-1)). No activation differences related to the task were observed during block 3 (propofol effect-site concentration: 4.35 mug ml(-1)). Conclusion. Our findings suggest sequential effects of propofol on auditory language processing networks. Brain activation firstly declines in the frontal lobe before it disappears in the temporal lobe.