We conducted a retrospective parallel longitudinal cohort study comparing surgical and medical treatment for epilepsy. The surgical group contained 201 patients treated with resective surgery for epilepsy in Norway since the first operation in 1949 until January 1988. The 185 control group patients treated medically only were closely matched for year of treatment, age at treatment, sex, seizure type, and neurologic deficit before treatment. There was no significant difference in survival between the two groups. The total monthly seizure frequency in the first and second year after operation and last year of registration (median 9 years) was significantly lower in the surgical group than in the control group (Mann-Whitney U test, two-tailed p less than 0.0001). The patterns were similar, with significant differences for subgroups with similar pretreatment status, such as seizure frequency, age, etiology and EEG-focality. Twenty-three and four-tenths percent (n = 40) of the surgically treated, and 2.9% of the controls had contracted neurologic deficits within 2 years after treatment. The difference was significant (chi square = 32.89, p less than 0.0001). Psychosis or permanent psychotic symptoms were reported in 6.7% (n = 11) of the surgically treated patients, and we suspect a higher proportion of psychotic development in the surgical group than in the control group. We conclude that surgical treatment for partial epilepsy is more successful than medical treatment in producing seizure reduction, provided the indications for operation exist. Surgical treatment produces more neurologic deficits than medical treatment (and possibly more psychiatric morbidity), and this factor must be weighed against the reduction in seizure frequency. The two treatments are equal for longterm survival.