Sixteen patients suffering from cluster headache received an injection of 160 mg methylprednisolone into the region of the greater occipital nerve ipsilateral to the pain, when free of headache. In episodic cluster (n = 8), attacks ceased in 2 cases and decreased in severity and frequency in one other. Three patients failed to respond at all. Two became headache-free, but the injection had possibly been performed towards the time when the episode might have been expected to end spontaneously. In chronic cluster (n = 8), four patients failed to respond at all. Four others improved, but only partially and transiently. On the whole, the present results are not as favourable as those obtained by Anthony, who advocated this technique.