In recent years, many attempts have been made to estimate shaft crack specifications with the least possible error. The presence of crack changes the physical characteristics of a shaft which in turn alter its dynamic response characteristics. Number of cracks, Crack depth and location are the main parameters influencing the vibration characteristics of the rotating shaft. In the present study, a technique based on the measurement of change of natural frequencies has been employed to detect the multiple cracks in rotating shaft. The model of shaft is generated using Finite Element Method for analysis. In Finite Element Analysis, the natural frequency of the shaft is calculated by modal analysis using the software ANSYS. The Numerical data are obtained from FEA, then used to train through Adaptive Neuro-Fuzzy- Inference System. Then simulations are carried out to test the performance and accuracy of the trained networks. The simulation results show that the proposed ANFIS estimatethe locations and depth of cracks precisely.