An initial study of the use of open access exercise electrocardiography by general practitioners (GPs) in South East Kent showed that patient selection and interpretation of test results was frequently incorrect. After issuing guidelines, modifying the request form and instituting registrar review of all requests, significant improvements in both referral pattern, result interpretation and patient management have resulted. Forty-nine GPs requested 110 exercise tests during 1988/89. Twelve per cent were not undertaken after discussion with the referring practitioner. Eighty-four per cent of those tested would have been referred to the district general hospital cardiology outpatient department in the absence of open access exercise electrocardiography service. Six per cent of patients were referred directly for invasive investigation. Thirty-five per cent were referred to the district general hospital cardiology outpatient department, whilst 42% were spared hospital referral based on the result of the investigation. Better use of the modified service was suggested by: referral of fewer patients with non-cardiac chest pains (P = 0.002); more patients with a moderate pre-test probability of ischaemic heart disease (P less than 0.05); fewer inappropriate requests (P less than 0.01); and fewer inappropriately undertaken tests (P less than 0.001) than in the previous study. All patients with strongly positive test results were appropriately managed. Open access exercise electrocardiography in the format investigated is potentially a cost-effective and useful tool to improve diagnosis and management of heart disease by GPs.