OBJECTIVE--To assess general practitioners' decisions to request lumbar spine radiographs according to the guidelines of the Royal College of Radiologists. DESIGN--Prospective questionnaire survey of outpatients attending for lumbar spine radiography. SETTING--London community hospital. SUBJECTS--100 consecutive adult outpatients attending for lumbar spine radiography at their general practitioner's request. MAIN OUTCOME MEASURES--Patient's history and clinical signs; radiological diagnosis; change in management of patients with significant radiological abnormality in response to the radiologist's report. RESULTS--60 patients were aged between 18 and 45, 27 (45%) of whom were women. Five patients were fully examined by their doctor before radiographs were requested, 76 were partially examined, and 19 were not examined. In 37 patients the examinations showed radiologically normal findings; 30 had radiologically significant disc or degenerative disease. Pain score and radiological diagnosis was not correlated (6.43 (range 1-10) for patients with significant disease v 6.14 (range 1-10) for those without, p greater than 0.05). There were no cases of malignancy or infection. One patient with radiologically significant disease was referred to a hospital specialist, and the management of only two such patients was altered by the report. 52 of the examinations should not have been requested if the guidelines had been strictly applied. CONCLUSIONS--There is a need to inform doctors of the efficacy of radiological examinations. An awareness of the college's guidelines among general practitioners should be actively promoted by radiologists.