In this paper, we present an efficient lossless ECG compression method for real-time applications. The proposed method hybridizes ECG predictive coding algorithm based on R-wave identification and Takagi-Sugeno fuzzy neural network. The ECG signal is predicted and encoded by using the periodicity of the electrocardiogram, the correlation between the electrocardiogram signal and the adjacent heartbeat, as well as the continuous characteristic of the ECG waveform. By recording the original prediction errors, non-distortion compression of the ECG signal can be achieved. In the experiment, we used MIT-BIH arrhythmia ECG database as the test data. According to the experimental results, the proposed algorithm can have a compression ratio up to 3.25 in average, which justifies the usefulness of the proposed approach.