Thirty-five hydronephrotic kidneys, in which diuresis renography showed either an equivocal (13) or non-obstructive (22) response pattern, were studied by a modified method in which intravenous frusemide was given 15 min before the start of the renogram so as to assess elimination during the period of maximum diuresis. Thirteen kidneys were identified as obstructed, including three in which the standard diuretic renogram appeared to exclude obstruction. These results indicate that the modified method increases the specificity of diuresis renography when used to assess patients with equivocal pelviureteric junction obstruction.