Among the 4129 patients of the Community Hospital-based Stroke Program, 30% had an unspecified stroke diagnosis. Since specific diagnosis and, perhaps, eventual treatment, derives in part from diagnostic testing, we examined the effect of clinical condition, geographic and demographic factors on the incidence of certain diagnostic tests after acute stroke. In this multivariable analysis, race, sex, history of hypertension and history of diabetes did not influence the chance of having any test, but older age strongly reduced the chances of receiving extensive evaluation. When CT scanning was available, the utilization of a CT as well as other diagnostic studies including cerebral angiography, radionuclide brain scan, EEG and EKG was increased. The odds of receiving a CT scan increased if the patient was married, and decreased with a history of previous stroke. A history of previous TIA increased the chance of having a cerebral angiogram while a history of cardiac disease decreased the chance. There were striking regional geographic differences in the use of CT, radionuclide brain scanning and cerebral angiography which may, in part, reflect differences between the availability of these technologies in urban and rural hospitals. These results indicate that evaluation of stroke patients remains heterogenous. © 1991.