A method for classification of ECG beats into normal and different categories of abnormal beats. The method comprises the following steps, performed on a computing platform: a. performing a training phase comprising taking as an input one or more ECG beats that are pre-classified into the different categories, with each ECG beat decomposed into multiple features and defining membership functions for each beat category for each feature; b. performing an operating phase, wherein (i) for each ECG beat to be classified, decompose the ECG beat into the multiple features; (ii) compute values of each membership function identified in the step of performing the training phase for each feature of the ECG beat for the different categories; (iii) merge the membership function values for all features in an ECG beat in a manner as to get a single beat value for each beat category across the features; (iv) identify a value alpha in the range [0,1], such that the difference between the top two beat values across the different beat categories is more than or equal to the sum of the beat values of all the remaining beat categories multiplied by the value alpha and such that a desired percentage of normal or abnormal beat classification is obtained; (v) if an alpha value can be identified, then classify the beat to belong to the beat that has the highest beat value, otherwise classify the beat as abnormal.