The purpose of the study was to compare the positive and negative predictive values of conventional renography (Reno-A), captopril renography (Reno-B) and ultrasound Doppler (UD) with regard to the diagnosis renal artery stenosis. These three tests, and in addition a renal angiography, were performed in consecutively admitted patients with arterial hypertension, owing to either suspicion of renovascular hypertension or refractoriness to treatment. Patients with occlusion of a renal artery or a serum creatinine level higher than 300 mumol/l, or a previous investigation for renovascular hypertension at another hospital, were excluded from the analysis. The European Multicenter Study (EMS) criteria and local criteria for abnormal renography were compared. Of 131 patients, 28 had a renal artery stenosis (RAS) exceeding 50% reduction in diameter of the artery and 19 exceeding 70%. Using the EMS criteria for renography the predictive values of a negative test for a RAS more than 50% were 0.88 for Reno-A, 0.90 for Reno-B, 0.86 for changes from Reno-A to Reno-B, 0.92 for abnormalities either in Reno-A, Reno-B or changes from Reno-A to Reno-B, and 0.91 for UD. The corresponding values for a RAS more than 70% were 0.94, 0.97, 0.93, 0.98 and 0.96, respectively. The predictive values of a positive test were clearly lower, ranging from 0.20 to 0.75, but best when changes from Reno-A to Reno-B were used, 0.69-0.75. Using local criteria for renography the predictive values of a negative test were almost equal to those obtained by using the EMS criteria, but the predictive values of a positive test were slightly lower. It is concluded that conventional renography, captopril renography and ultrasound Doppler all are very good screening tests for renal artery stenosis, but the positive predictive values are clearly highest when using changes from conventional renography to captopril renography. It is suggested that captopril renography always should be performed when conventional renography is abnormal and vice versa to obtain the highest positive predictive value, on the assumption that total renal function is normal or almost normal, and that renal function is not absent in the affected kidney.