Warfarin dosing is challenging due to a multitude of factors affecting its pharmacokinetics (PK) and pharmacodynamics (PD). A novel personalised dosing algorithm predicated on a warfarin PK/PD model and incorporating CYP2C9 and VKORC1 genotype information has been developed for children. The present prospective, observational study aimed to compare the model with conventional weight-based dosing. The study involved two groups of children post-cardiac surgery: Group 1 were warfarin naïve, in whom loading and maintenance doses were estimated using the model over a 6-month duration and compared to historical case-matched controls. Group 2 were already established on maintenance therapy and randomised into a crossover study comparing the model with conventional maintenance dosing, over a 12-month period. Five patients enrolled in Group 1. Compared to the control group, the median time to achieve the first therapeutic INR was longer (5 vs. 2 days), to stable anticoagulation was shorter (29.0 vs. 96.5 days), to over-anticoagulation was longer (15.0 vs. 4.0 days). In addition, median percentage of INRs within the target range (%ITR) and percentage of time in therapeutic range (%TTR) was higher; 70% versus 47.4% and 83.4% versus 62.3%, respectively. Group 2 included 26 patients. No significant differences in INR control were found between model and conventional dosing phases; mean %ITR was 68.82% versus 67.9% ( p = 0.84) and mean %TTR was 85.47% versus 80.2% ( p = 0.09), respectively. The results suggest model-based dosing can improve anticoagulation control, particularly when initiating and stabilising warfarin dosing. Larger studies are needed to confirm these findings.