Abstract Introduction: Determining renal function (i.e glomerular filtration rate) is in many ways essential in clinical practice, in terms of diagnosis and drug therapy. Exogenous markers, such as iohexol, are mainly used for a more accurate determination of GFR. A low dosage of iohexol is injected to the patient and GFR is determined by using an algorithm and plasma concentrations measured at 2 and 5 hours after administration. Non-parametric population pharmacokinetic models are invaluable tools for describing the pharmacokinetic properties of different drug and are suitable for identifying subpopulations deviating pharmacokinetic properties. Population models, unlike algorithms are not dependent on specific sampling times. Method: A standardized dose of iohexol is injected and up to 12 plasma concentrations of iohexol are measured with HPLC-UV. The pharmacokinetic properties are investigated using non-parametric population modeling (Pmetrics®). With the help of the MMopt function in Pmetrics®, the most informative sampling times for determining the AUC of iohexol are identified. The developed model is then utilized to investigate optimal sampling times to determine the AUC for iohexol, which is then used to determine individual GFR. The model is assessed against AUC determined with the trapezoidal method and also the standard method where GFR is determined using plasma concentrations 2 and 5 hours after administration. Results: 13 renal transplant recipients were included in the study. A 2-compartment model with primary parameters allometrically scaled with centralized body weight described the pharmacokinetic properties of iohexol best. Individual observation-prediction plot showed r2-values at 0.996. With the optimal sampling times determined with the help of the MMopt function, GFR was determined with both 2-and 3-point measurements. Predicted GFR showed close values to the reference for most but was massively underpredicted for certain patients. Conclusion: The massive underprediction was most likely due to the limited number of test subjects included in this study. Future work on building a better model will require the inclusion of more patients; especially more data in the later post administration phase in patients with low renal function.