Lumbar facet disease is sometimes implicated in low back pain. Identification is difficult and this may account for a variable response. Single photon emission computerized tomography (SPECT) is a scanning technique which enables localization of facet joint pathology. We determined whether recognition of facet disease by this method improved the response to corticosteroid injection treatment. Fifty-eight patients with low back pain and displaying accepted clinical criteria for facet joint disease were evaluated by SPECT. Twenty-two had facetal uptake of isotope. These and the tender facet joints of 36 scan-negative patients were injected with 40 mg methylprednisolone and 1 ml 1% lignocaine under X-ray control. Pain was assessed by a blind observer using the McGill questionnaire (MGQ), Present Pain Intensity score (PPI) and a Visual Analogue Scale (VAS). VAS, PPI and MGQ were reduced in the scan-positive patients at 1 month (P = 0.05, P = 0.0005, P = 0.005) and MGQ at 3 months (P = 0.01), whilst scan-negative patients were unchanged. The percentage of scan-positive patients who reported improvement was 95% at 1 month and 79% at 3 months, significantly greater than the control group (P = 0.0005, P = 0.01). Within 6 months, pain improvement in the SPECT-positive group was no longer statistically significant. Tenderness did not correlate with increased uptake on SPECT scan. Osteoarthritis of the facets was more common in the SPECT-positive patients (P < 0.001), but did not correspond with sites of increased uptake on SPECT scan. These results suggest that SPECT can enhance the identification of back pain sufferers likely to obtain short-term benefit from facet joint injection.