The purpose of the present study was to compare different pharmacokinetic models for estimation of glomerular filtration rate (GFR) in 50 dogs with pyometra. GFR was estimated by plasma clearance (CLplasma) of iohexol by four 1-compartment methods (CL1c), a 2-compartment method (CL2c), and the trapezoidal method (CLtr). Regression analysis was performed to establish correction formulas for prediction of CLtr from the CL1c values and to find optimal times of sampling. Standardization of clearance values to body weight (kg), body surface area (m2) and extracellular fluid volume (ECFV) was compared by ranking of values. CLtr and CL2c values were similar, whereas CL1c overestimated CLtr. CLtr could be predicted from 2 samples at 2 and 3 hours after injection, using the formula CLtr = 4.52 + 0.84CL1c - 0.00080(CL1c)2 (R2 = .97). Similar relationships were found when sampling at 2 and 4 hours or at 2, 3 and 4 hours after injection, whereas predictions from the 3- and 4-hour estimates were not optimal (R2 = .79). The 2-sample methods for calculating GFR/ECFV generally produced unreliable predictions of the complete curve GFR/ECFV values. For some dogs, the choice of standardization procedure substantially changed the apparent level of renal function relative to other dogs in the study. We conclude that by applying an appropriate correction formula, GFR may be estimated using 2 blood samples at 2 and 3, or 2 and 4 hours after injection of iohexol when renal function is normal or moderately reduced. The method of standardizing the analysis with respect to body size may influence interpretation of the results substantially.