Bone milling is one of the most important and sensitive biomechanical processes in the field of medical engineering. This process is used in orthopedic surgery, dentistry, treatment of fractures, and bone biopsy. The use of automatic numerical control surgical milling machines has revolutionized this procedure. The most important possible complication in bone surgery is the rise of temperature above permissible range and the formation of thermal necrosis or cell death in bone tissue. In the present article, a study on the design of experiment is first conducted by considering the rotational speed of the utilized tool, feed rate, depth of cut and tool diameter as the most important input factors of this process. Then, an adaptive neuro-fuzzy inference system (ANFIS) is developed to model and estimate the temperature behavior in the process of robotic bone milling. The optimal parameters of the ANFIS system are obtained using teaching-learning-based optimization (TLBO) algorithm. In order to model the process behavior, the results of experiments are used for the training (75% of the data) and testing (25% of the data) of the adaptive inference system. The accuracy of the obtained model is investigated via different plots, and statistical criteria, including root mean square error, correlation coefficient, and mean absolute percentage error. The findings show that the ANFIS network successfully predicts the temperature in the automatic bone milling process. In addition, the network error in estimating the temperature of the automatic bone milling process in the training and test section is equal to 1.74% and 3.17%, respectively.