The glomerular filtration rate (GFR) is used clinically to assess renal function. The most accurate estimation technique is tracer clearance where deterministic compartment pharmacokinetic models are most widely used. The aim of this study was to assess the viability of alternative pharmacokinetic models to describe tracer clearance, and in turn, measure GFR. This study was carried out on 126 clearance datasets obtained from 44 patients with large solid tumours; these were fitted to four pharmacokinetic models with superiority of model determined by Akaike Information Criteria. A fractal model was found to be superior to the best deterministic compartment model (70% of datasets, P < 0.0020) as was a gamma-distributed residence time model (93% of datasets, P < 0.0020); both models also gave greater mean weighted coefficients of determination than deterministic compartment models. These results suggest that gamma-distributed residence time and fractal models better describe tracer clearance than deterministic compartment models and therefore should allow more accurate estimation of GFR.