Effective surgical treatment of patients with intractable complex partial seizures depends on accurate preoperative seizure focus localization. We evaluated seizure localization with interictal and immediate postictal single photon emission computed tomographic images of cerebral perfusion using technetium-99m-hexamethyl-propyleneamineoxime (99mTc-HMPAO) in comparison with conventional ictal electroencephalographic (EEG) localization. Thirty-two patients with intractable complex partial seizures were studied. The mean delay from seizure onset to injection was 6.3 +/- 5.3 (SD) minutes. Independent blinded observers assessed the scans for interictal hypoperfusion and postictal focal hyperperfusion. Interictal scans alone were unreliable, indicating the correct localization in 17 patients (53%) and an incorrect site in 3 (9%). When interictal and postictal scans were interpreted together, the focus was correctly localized in 23 patients (72%). There was 1 false-positive study, and 8 patients had inconclusive changes, including 2 with inconclusive depth EEG studies. Postictal hyperperfusion was predominantly mesial temporal and frequently associated with hypoperfusion of lateral temporal cortex. Secondarily generalized seizures tended to show focal hyperperfusion less often than complex partial seizures did (Fisher's exact test p = 0.09). Combined interictal and immediate postictal single photon emission computed tomography with 99mTc-HMPAO is a useful noninvasive technique for independent confirmation of electrographic seizure localization. It may provide a suitable alternative to the use of depth electrode studies for confirmation of surface EEG findings in many patients with complex partial seizures.