In total knee arthroplasty (TKA), real time optical measurements are now carried out to improve the implantation of prostheses. The final alignment of prostheses is however influenced by the choice of points probed by the surgeon, and by the measurement accuracy. A Monte Carlo method has been developed, to simulate a whole set of measurements. To estimate the uncertainties of the resultant angles between the two parts of the knee prosthesis, this data has then been treated with the algorithms implemented in Computer Assisted Orthopaedic Surgery programs. Such results were also obtained through uncertainty propagations based on the GUM.