A method of estimating subject specific muscle parameters of musculoskeletal models of elite athletes (skiers) was sought. Subject specific models are necessary due to large differences in general anatomy and physical performance of elite athletes relative the general population. Sought muscle parameters concern the force generating capabilities of muscles. The estimation was limited to only include the quadriceps-femoris and hamstring muscle groups due to these muscle having the highest influence on the performance of a skier. A modified interpretation of the method proposed by Heinen et al. [19] was implemented. The method includes experimental strength tests of knee extension and flexion muscles of a test subject, a musculoskeletal model of the experiments coupled with a mathematical optimisation minimisation formulation. The aim of the optimisation was to match the strength of a model to the experimentally obtained strength curve by minimising  error between the model and experimental results. The optimisation minimises the error between the model and the experimental data by varying the operating range and strength of the involved muscles. The musculo-tendon parameters are estimated through transformation equations, explicitly related to the design variables. Three healty and active males were involved in this study. An overall increase of the accuracy of the optimised model relative an unscaled reference model was observed, with the reduction of the objective function in a range of 80.2-92\% and a mean absolute error varying between 6.8 to 16.5 Nm. In the case of quadriceps-femoris muscles, the optimised model struggles with incorrect prediction of the peak torque and peak torque angle due to limitations of the muscle model and the distribution of the moment arm. The model predicts both peak torque and peak torque angle with high accuracy in the case of hamstring muscles. In addition, the model struggles with low precision for both knee extension and flexion for all of the involved test subjects. Although great improvement in the accuracy was observed, the model prediction was deemed to have low clinical significance, due too low accuracy and precision. The clinical significance could be improved, for example by a more detailed musculoskeletal model or by modifying the behaviour of the muscle model. Future work should focus on addressing the current issues presented in this study and a further development, as the method still is relatively new and untested. Parallelly, the researches should try to test the method in clinical studies, in order to evaluate the influence on the results by the implementation of this method of parameter estimation.