Fibre-optic bronchoscopy is widely used to diagnose bronchial carcinoma. There is considerable variation in techniques for patient sedation, methods of obtaining samples and histopathological yield. We wished to examine variations in practice in different centres throughout Scotland and derive realistic audit standards for best clinical practice from these results. Diagnostic bronchoscopies from five centres were included. Patient details, grade of individual performing the test, endobronchial abnormalities, specimens taken and the histocytological yield were recorded. A patient satisfaction questionnaire was completed. One thousand eight hundred and two bronchoscopies were performed to look for bronchial carcinoma. Sedation and anaesthesia techniques varied considerably between centres. There were marked differences in patient satisfaction between centres. Nearly twice as many females as males would prefer not to have bronchoscopy repeated. Six hundred and fifty-eight carcinomas were confirmed by histocytology. Yield was unaffected by the grade of doctor performing bronchoscopy. Improving yield may be achieved by increasing the number of sampling techniques employed and changing the order in which specimens are taken (biopsies first and washings last). Eighty-seven percent of endoscopically visible tumours were confirmed histocytologically. There was a considerable variation in histological spectra between centres that may relate to differences in pathological interpretation rather than actual differences in case mix. Suggested audit standards are discussed. This study demonstrates the variety of techniques and also the levels of histocytological yield and patient satisfaction that can be achieved. Provisional standards of practice for this procedure have been agreed with a view to auditing performance against these. It is hoped that centres will adopt the methods that are shown to achieve the highest standards.