The aim of this work was to develop an online simulator for levodoopa infusion in Parkinson’s disease based on a pharmacokinetic model. Using this simulator, optimal infusions setting were calculated, the relation between flow rate and morning bolus dose and usefulness for optimising the titration procedure of new infusion patients based on cased-based reasoning were investigated. Results show that the simulator was quite useful for simulating medication treatment and the relationship between optimal flow rate and morning dose seemed linear with some reasonable outliers. The prediction error revealed a tendency of simulator to under predict the actual flow rates and to over predict morning doses. At last, the simulator predicted well on each single occasion of majority of patients, but different flow rates and morning doses on different occasions of the same patient resulted in different matching parameters in database.