A computerized analysis of EEg-records from 96 multiple sclerosis (MS) patients yielded about twice as many abnormal results (79%) in comparison to the usual visual EEG-evaluation (39%). The statistically significant (p less than 0.05) increase in slow frequencies in addition to beta frequencies to the cost of alpha, but especially the reduction in total percentage of alpha contribution to overall EEG-power discriminated most efficiently between MS-patients and our normal population. Thus, alpha index alone allowed for 87% correct classification of the normal population, and 73% of patients with normal visual EEG-records were correctly classified as belonging to the MS-population. Extending the patient group to include pathological visual EEG-records as well, the corresponding figures were 95% and 89%, respectively. The computerized EEG-analysis was thus clearly superior to the usual visual inspection.