Centrifugal pump is one type of pumps that widely used, especialy in industry. It’s mechanism which cretates pressure changes usually caused cavitation. Cavitation phenomenon that is not properly maintain results fatal breakdown and high economic losses. Therefore, research is needed to find and develop the method that can detect early cavitation phenomena in centrifugal pumps, and can show cavitation at several levels. This paper presents a method that able to detect cavitation by monitoring the vibrations level of the pump based on statistical analysis of time domain and Principal Component Analysis (PCA). By using Matlab software, data is trained and tested in each condition. Training data is normalized and trained from each condition using PCA and will produce data loading matrix. After that, the loading matrix is multiplied by the testing data in each condition so that it produces a score that is used to classify the damage to the centrifugal pump. The result shows that the method of domain-based PCA proposed is successful in transforming 3500 data set from 7 statistical parameters to provide the 7 principal component (PC) with maximum variant. The identification accuracy shows 93.68% variants, PCA is able to clearly identify and classify the differences between normal, early cavitation, intermediate cavitation and advanced cavitation conditions in centrifugal pumps.