The aim of this study was to evaluate the variability of the Patlak-Rutland slope as applied to the renal clearance method. We analysed the 99Tcm-mercaptoacetyltriglycine renograms of 17 non-selected patients (both children and adults, number of kidneys = 34) with a single kidney glomerular filtration rate of 6-73 ml min-1 1.73 m-2. the acquisition time for the renograms was 20 min with a frame duration of 20 s. First, correction for background activity was introduced using the perirenal area and the size ratio between the kidney and the perirenal area. The Patlak plot was then applied. The first two points were systematically rejected and a series of linear fits calculated from the experimental points, starting from the third point (i.e. 1 min). The final point of the fit never exceeded the time of the peak of the renogram minus one minute (Tmax - 1) and was always less than 5 min. Although simple visual inspection of the slopes would suggest that the experimental points were distributed accurately along a straight line, it appeared that the slope was strongly dependent on the number of points used for the determination of the fit. As a second step, we restricted the number of points for fitting to the time interval between 1 min and 2 min 40 s. This procedure appeared to show that the Patlak methodology is reliable, giving rise to only small variations in slope, depending on the number of points chosen. However, even in this narrow time interval, significant errors can be made, either because of the very early escape of the radionuclide or because of statistical noise (e.g. in renal failure the signal-to-noise ratio is unfavourable). Therefore, it is advisable, when using this methodology, to restrict the fitting procedure to the second minute of the study and to check visually that this fixed time interval gives rise to a slope that is well adapted to the Patlak plot.