This study evaluates the specificity of facet joint injections in diagnosing symptomatic facet joint pain. Fifty-four patients had 143 facet joint injections performed in a prospective study. Pain provocation and pain relief were used to confirm the diagnosis. All patients had symptoms and signs considered suggestive of facet joint pain. Less than 1.5 ml of bupivacaine was injected into each joint to prevent extravasation. Only nine patients (16.7%) had the diagnosis confirmed, significantly lower than most published results. An anatomic study performed suggests that extravasation into the epidural space occurs following rupture of the joint capsule. This provides an explanation by which good therapeutic results can be obtained if large volumes of therapeutic agent are used.