One hundred and seventy-six consecutive patients with chronic low back pain and no history of previous lumbar surgery were studied to determine the false-positive rate of single diagnostic blocks of the lumbar zygapophysial joints. All patients underwent diagnostic blocks using lignocaine. Those patients who obtained definite or complete relief from these blocks subsequently underwent confirmatory blocks using bupivacaine. Eighty-three patients (47%) had a definite or greater response to the initial, lignocaine injection at one or more levels but only 26 (15%) had a 50% or greater response to a confirmatory injection of 0.5% bupivacaine. Using the response to confirmatory blocks as the criterion standard, the false-positive rate of uncontrolled diagnostic blocks was 38% and the positive predictive value of these blocks was only 31%. Because the positive predictive value of a test is lower when the pre-test probability (prevalence) is low, and because the prevalence of lumbar zygapophysial joint pain is likely to be less than 50%, uncontrolled diagnostic blocks will always be associated with an unacceptably low positive predictive value. These features render uncontrolled diagnostic blocks unreliable for the diagnosis of lumbar zygapophysial joint pain not only in epidemiologic studies but also in any given patient.