Twenty-five patients with occipital lobe seizure origin were retrospectively evaluated to determine clinical seizure characteristics and electroencephalographic manifestations. Certain symptoms and signs served to identify occipital lobe origin in 22 (88%). These included elementary visual hallucinations, ictal amaurosis, eye movement sensations, early forced blinking or eyelid flutter, and visual field deficits. Eye or head deviation, or both, was observed frequently and was contralateral to the side of seizure origin in 13, but 3 patients exhibited ipsilateral deviation in some or all their seizures. After the initial signs and symptoms, clinical seizure characteristics resembled those of seizures originating elsewhere. Seizures typical of temporal lobe origin with loss of contact and various types of automatic, semipurposeful activity occurred in 11 patients. Seizures in 3 patients exhibited asymmetrical tonic or focal clonic motor patterns characteristic of frontal lobe seizures. Eleven of the 25 patients had, on two occasions, two or more distinctly different seizure types. Scalp electroencephalographic findings were seldom helpful for occipital lobe localization and were frequently misleading. Intracranial electroencephalographic recording correctly identified occipital lobe seizure origin in most, but not all, patients who had such studies. Intracranial electroencephalic recording also proved the variability in clinical seizure characteristics was related to different seizure spread patterns, medially or laterally above and below the sylvian fissure, both ipsilateral and contralateral to the occipital lobe of seizure origin. Eighteen patients had occipital lobe lesions detected with computed tomographic or magnetic resonance imaging scans or both. Resection of the lesions in 16 patients produced excellent results in 14 (88%). Five patients had temporal lobectomies, with good results in 3, but poor results in 2. Two patients with unlocalized seizures had complete section of the corpus callosum, 1 with a good result and the other with a poor result.