OBJECTIVES To determine the number of inappropriate requests for electroencephalography (EEG) and whether guidelines on use could reduce this number. DESIGN Audit with retrospective and prospective components. SETTING EEG department in district general hospital and centre for neurology and neurosurgery. PARTICIPANTS Retrospective: 368 at the general hospital and 143 patients at the neurology centre. Prospective: 241 patients undergoing EEG at the general hospital. INTERVENTIONS Guidelines for EEG issued to users of service at the general hospital. OUTCOMES Retrospective: differences in requesting practice, result in different clinical scenarios, relative roles of procedure, clinical acumen in establishing diagnosis, usefulness of procedure. Prospective: change of requesting practice, impact on use. RESULTS There were considerable differences in requesting practice. Non-specialists seem to use EEG as a diagnostic tool, especially in patients with "funny turns," when it is much more likely to yield potentially misleading than clinically useful information. The overall proportion of procedures considered to influence management, to be justifiable, and to be inappropriate were 16% (59), 28.3% (104), and 55.7% (205), respectively. In the prospective study the total number of requests was significantly reduced (chi(2)=33.85, df=5, P<0.0001), mainly because of fewer requests in patients with non-specific "funny turns" (chi(2)=21.90, df=6, P=0.0013). There was a concomitant change in the usefulness of EEG (chi(2 )=26.99, df=2, P<0.0001). CONCLUSIONS This original audit informed clinical practice and had potential benefits for patients, clinicians, and provision of service. Systematic replication of this project, possibly on a regional basis, could result in financial savings, which would allow development of accessible local neurophysiology services.