The postoperative magnetic resonance imaging studies of 40 patients who underwent surgery for medically refractory temporal lobe epilepsy were reviewed. A method for quantifying the resection of four specific medial temporal lobe structures--amygdala, uncus, hippocampal formation, and parahippocampal gyrus--was used to correlate postoperative seizure control with the degree to which those structures had been resected. Satisfactory postoperative seizure control was found to be strongly dependent (P less than .01) on the performance of at least partial resection of all four medial temporal lobe structures. The total amount of tissue removed did not necessarily correlate with the clinical outcome in individual cases, however, as long as some portion of all four medial structures had been resected. Two complicating factors--the presence of posterior or bilateral independent electroencephalographic foci and a history of meningoencephalitis--each negatively influenced surgical outcome (P less than .01). In cases with a complicating factor and inadequate medial resection, the outcome was invariably poor.