Two methods of glomerular filtration rate estimation have been evaluated, based on the intravenous administration of 99mTc-DTPA and the measurement of renal time activity curves by means of a computer linked gamma camera. A single 20 min plasma sample was also required. These methods were designed to minimize the component of error arising from decay statistics. One method was based on using a constant fraction of the cardiac activity in lieu of a perirenal region of interest for the background correction, the other was based on deconvolution by a constrained least squares technique. The first method, based on modifying the background correction, led to poor results (residual standard deviation 18.9 ml/min when compared with the plasma clearance method). The second method, based on constrained least squares deconvolution, worked as well as previously reported methods (residual standard deviation 14.5 ml/min) and appears suitable for clinical use.