In clinical practice and in many reported studies about incidences and time courses of vasospasm the angiographic spasms are judged by eye without clear definition of vessel narrowings. To evaluate the reliability of this diagnostic method two experienced neuroradiologists and two experienced neurosurgeons independently in two sessions, examined 30 carotid angiograms performed after an aneurysmal subarachnoid haemorrhage. The intra- and inter-observer agreements for the absent/present and localization diagnosis of vasospasm were calculated by means of Kappa statistics. Kappa values for both intra- and inter-observer agreement showed great variability and in general most of the agreements were not much better than chance expected agreement. The diagnostic method of judging angiographic vasospasm by eye without clear-cut definitions of vessels narrowings is unreliable and should not be used in the future, neither in clinical practice nor in research.