In order to improve the quality of service and limit costs of a local electromyographic service, the requests for electromyographic (EMG) examination issued by general practitioners and specialists were critically evaluated. The diagnoses suspected by referring doctors were compared with medical history and clinical data to analyze (1) the utility of EMG for the final diagnosis and (2) the concordance with that of the neurophysiologist and with the EMG results. In 1994 and 1995, there were 3482 patients referred to the EMG service. Only patients undergoing EMG for the first time (2706 subjects) were considered. In 76.6% of cases, the referring doctor had indicated the suspected diagnosis in the request. This diagnosis was compatible with medical history, symptoms and the results of neurological examination in 57.6% of cases. In 77.2%, the neurophysiologist considered the EMG useful in confirming the clinical diagnosis. The suspected diagnosis was confirmed by the clinical diagnosis of the neurophysiologist and by the EMG results in 54.2% and 45.4% of cases, respectively. When the request was issued by neurosurgeons, neurologists, orthopedists, rheumatologists and physiatrists, the suspected diagnosis was more accurate; as a consequence the EMG was more correctly oriented than when the request was issued by other specialists or by general practitioners. It is recommended that neurological examination be a prerequisite for EMG requests issued by general practitioners.