Most patients with primary headache syndromes who have frequent attacks of pain have tenderness in the sub-occipital region. Injection of the greater occipital nerve (GON) with local anesthetic and corticosteroids has been widely used in clinical practice for many years, yet there is no clear understanding of its mechanisms of action. Moreover, there is no current gold-standard of practice regarding GON injections in the management of headache. We audited of our practice to generate hypotheses about the range of primary headaches that might benefit, to determine response rates to power future studies, and to assess whether we should continue to do this procedure. Twenty-six of fifty-seven injections in 54 migraineurs yielded a complete or partial response that lasted for the partial response a median of 30 days. For cluster headache 13 of 22 injections yielded a complete or partial response lasting for a median of 21 days for the partial response. Tenderness over the GON was strongly predictive of outcome, although local anesthesia after the injection was not. The presence or absence of medication overuse did not predict outcome. Apart from two patients with a small patch of alopecia the injection was well tolerated. GON injection is a useful tool in some patients that provides interim relief while other approaches are explored. It is remarkable that in all conditions in which an effect is observed the response time so much exceeds the local anesthetic effect that the mechanism of action may well be through changes in brain nociceptive pathways.