The techniques, results, and problems of three types of selective temporal lobe (TL) amobarbital procedures (balloon technique with temporary occlusion of the internal carotid artery distal to the origin of the anterior choroidal artery (acha) [n = 19]; selective anterior catheterization of the acha [n = 20]; and selective catheterization of the peduncular P2-segment of the posterior cerebral artery [n = 5]) are described in a group of 40 patients with medically refractory complex partial seizures of mesial TL origin. Selective amobarbital tests were carried out before surgery to predict the memory deficit after an intended selective amygdalohippocampectomy. The effects of selective anaesthetization of TL were correlated with clinical data, pattern and duration of amobarbital induced EEG changes, and performance on verbal and nonverbal memory tasks measured during the test. In 4 patients the effect of selective amobarbital injection on regional and global metabolism was studied with 18F-FDG-PET, with the PET tracer being injected intravenously immediately after amobarbital. More recently in 2 patients the vascular territory perfused by amobarbital in the acha test was studied with SPECT using 99m Tc ECD injected immediately prior to the amobarbital into the acha. Whereas the PET studies showed a rather widespread and bilateral amobarbital-induced decrease of metabolism, the SPECT studies confirmed the selective distribution of the tracer in the vascular territory of the acha, i.e., in amygdala and hippocampus. The comparison of selective TL amobarbital test performance with postoperative neuropsychological performance showed that the predictive value of this test is rather good for the postoperative verbal memory but underestimates postoperative nonverbal ("figural") memory performance.