The increased use of disc brakes in passenger cars has led the research world to focus on the prediction of brake performance and wear under different working conditions. A proper model of the brake linings’ coefficient of friction (BLCF) is important to monitor the brake operation and increase the performance of control systems such as ABS, TC and ESP by supplying an accurate estimate of the brake torque. The literature of the last decades is replete with semi-empirical and analytical friction models whose derivation comes from significant research that has been conducted into the direction of friction modelling of pin-disc couplings. On the contrary, just a few models have been developed and used for the prediction of the automotive BLCF without obtaining satisfactory results. The present work aims at collecting the current state of art of the estimation techniques for the BLCF, with special attention to the models for automotive brakes. Moreover, the work proposes a classification of the several existing approaches and discusses the relative pro and cons. Finally, based on evidence of the limitations of the model-based approach and the potentialities of the neural networks, the authors propose a new state observer for BLCF estimation as a promising solution among the supporting tools of the control engineering.