Biometric systems have been a subject of considerable research interest for human identification. In this paper, we present both fiducial and non-fiducial methods for individual identification. In both methods, we randomize the process of data selection, and we use different referencing and testing windows to examine the stability of our methods. Our identification system randomly selects both the referencing and testing data by utilizing multiple data windows from the full ECG record. The first identification method is based on extracting multiple bivariate histograms of fiducial QRS features. Namely, these features are the amplitude and slope differences between the Q, R and S peaks. Then, we find the Euclidean distance between these multiple histograms for classification purposes. Additionally, we propose an algorithm which automatically locates and segments QRS waves using Short Time Fourier Transform (STFT) and single feature-based classification process. The second non-fiducial identification method is based on finding the magnitudes of the frequency components from the ECG data