In recent years there has been a tendency to abandon repeat cerebral angiography in patients with subarachnoid hemorrhage(SAH) if the initial angiogram is normal because prognosis is said to be excellent. Our experiences does not support such a conclusion. In 291 cases with subarachnoid hemorrhage(SAH), four vessel cerebral angiography was performed to investigate a source of the hemorrhage. The first four vessel study on admission revealed 234 cases(80.4%) of ruptured cerebral aneurysms. Fifty-five of 291 patients with primary subarachnoid hemorrhage(SAH) did not show vascular lesion in initial pan-angiography studies. In order to define the benefit of control angiography these patients were reviewed. Among 41 patients who underwent repeat angiography, a second angiography showed an aneurysm in 24(58.5%). Twelve of the twenty-four cases had aneurysms in the anterior communicating artery. But the age, sex, initial clinical grade and CT grade of patients with aneurysm in repeat angiography were indistinguishable from thoses of patients without aneurysm. Even in the patients who showed angiographic vasospasm in intial angiography, probability of revealing an aneurysm was not different from those without spasm. We conclude that repeat angiography is necessary in the cases of SAH with normal initial angiogram regardless of the CT findings and the presence of vasospasm.