An intensive programme of guideline development and dissemination is not always practical. We designed general practitioner guidelines for chest radiography and disseminated them to a group of 33 practices selected randomly from the practices that refer more than five patients for chest radiography annually. The subsequent 2500 requests were analysed and demographic data, the referring practice and the relevant history noted. We documented a reduction in the proportion of referrals that were contrary to the guidelines by 30.5% (from 8.2% to 5.7%, P = 0.016), a reduction in referral rate with an inadequate history by 29.7% (from 15.5% to 10.9%, P = 0.0008) and an increase in the frequency that a presumptive diagnosis was stated by 14.5% (from 42.9% to 49.1%, P = 0.0025). There was no overall reduction in referral rate in the study group and the reasons for this are discussed. We conclude that guidelines for chest radiography are effective in reducing inappropriate requests and can also improve the clinical information provided.