Abstract: The key to condition monitoring and fault diagnosis of mechanical equipment is the characterization and extraction of fault features. The multi-scale entropy and its related methods-based nonlinear dynamic indexes can be used to effectively extract the nonlinear fault characteristic information contained in vibration signals. Since the introduction of the entropy method, the accuracy of entropy estimation has been improved and the application areas of entropy have been expanded through modifications and improvements. Multiscale entropy further explores the complexity information contained in the other scales of the time series, which has been widely used in mechanical condition monitoring and fault diagnosis. In this paper, firstly the application of multi-scale entropy methods in mechanical intelligent fault diagnosis including single-scale entropy, multi-scale entropy, multi-scale fuzzy entropy, multi-scale permutation entropy and multi-scale scatter entropy was reviewed, meanwhile the advantages and disadvantages of the related methods were summaried. Aiming at the problem of multivariate data processing, the multivariate multiscale entropy extended from univariate to multivariate was reviewed. Finally, combining the problems and challenges faced by relevant methods for focused multiscale entropy in mechanical intelligent fault diagnosis, the three future development directions were outlooked, i.e., the application in industrial big data, the combination with the study of fault mechanism, and constructing entropy-based deep learning models from the interpretability perspective.