In modern medicine, knee replacement surgery is a major procedure and carries just the same risks as any surgery. Nowadays, the average (length of stay) LOS for knee replacement surgery can range from one to four days. In this paper, LOS of patients undergoing knee replacement surgery was evaluated using Multiple Linear Regression analysis. Various Machine Learning (ML) models - Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM) and Gradient Boosted Trees (GBT) - were implemented to predict this LOS. The obtained results were compared with those obtained from two other hospitals (University Hospital “San Giovanni di Dio e Ruggi d’Aragona” and those of the A.O.R.N. “Antonio Cardarelli”).