Temporary nerve blocks using local anesthetic are employed extensively in the evaluation of pain problems, particularly lumbosacral spine disease. Their specificity and sensitivity in localizing anatomic sources of pain have never been studied formally, however, and so their diagnostic and prognostic value is questionable. There have been anecdotal reports of relief of pain by temporary blocks directed to areas of pain referral, as opposed to areas of documented underlying pathology; but there has been no study to define the frequency or magnitude of this effect. We have examined the specificity and sensitivity of a battery of local anesthetic blocks in a series of 33 patients with a chief complaint of sciatica, attributable in all cases to spinal disease (radiculopathy, with some clinical features of arthropathy). As determined by blinded patient analog ratings in randomized sequence, three different nerve blocks were significantly more effective than control lumbar subcutaneous injection of an identical volume of 3 ml of 0.5% bupivacaine (P < 0.05). Not only paraspinal lumbosacral root blocks and medial branch posterior primary ramus blocks (at or proximal to the pathology), but also sciatic nerve blocks (distal or collateral to the pathology) produced temporary relief in a majority of patients. This confirmed the study hypothesis that false positive results are common, and specificity is low. For sciatic nerve blocks, specificity was between 24% and 36%. Patterns of responses specific to the established diagnosis of radiculopathy (i.e., root block most effective) had sensitivities between 9% and 42%. Statistical analysis of clinical and technical prognostic factors revealed that the only association with pain relief by any block were the effects of other blocks. The strongest association was between relief by sciatic nerve block and relief by medial branch posterior primary ramus (facet) block (P = 0.001, odds ratio 16.0). There were no associations between the results of blocks and clinical findings (history, physical examination, diagnostic imaging) in these patients, chosen for their homogeneous clinical presentation and absence of functional signs. Our findings indicate a limited role for uncontrolled local anesthetic blocks in the diagnostic evaluation of sciatica and referred pain syndromes in general. Negative blocks or a pattern of responses may have some predictive value, but isolated, positive blocks are non-specific. This lack of specificity may, however, be advantageous in therapeutic applications.