To evaluate the effect of brain retraction during neurosurgical procedures, changes of N(1) components of somatosensory evoked potential (SEP), cortical blood flow (CoBF) by a hydrogen clearance method and histology were studied following local brain compression with a brain spatula in 25 mongrel dogs. Local compression pressure (LCP) at 20 mmHg for 30 minutes was applied four times with five minute intermissions during which time the pressure was completely released. In the same way, the procedures were done at varying levels of LCP, namely 40, 60 and 80 mmHg respectively. Results were as follows: 1) The more the numbers of repetitive compression increased at each pressure level or the more LCP was raised, the more both N(1) amplitude and CoBF decreased. 2) When LCP was more than 40 mmHg and/or N(1) amplitude decreased by more than 50 % during brain compression, SEP did not recover for two hours after compression was ceased. In addition, histological study demonstrated cortical microhemorrhage and edema. It is likely that LCP over 40 mmHg and reduction of N(1) amplitude by more than 50 % during brain retraction produces postoperative neuronal dysfunction. 3) In the noncompressed hemisphere, N(1) amplitude, CoBF and histology were not changed to any significant degree.