A method based on The FNN is developed to create fuzzy membership functions for classification of cardiac arrhythmia in this paper. The FNN of Sugeno type is constructed firstly. The R_R interval and QRS complex are used as the inputs of the FNN. Then Cam Delta learning algorithm is used to train the FNN through which the membership functions can be gotten. The fuzzy recognition using these membership functions can classifg cardiac arrhythmia. The verification result shows that this method is effective.