In order to analyze arrhythmia, a noval practical method was proposed. Here the pRRx serials (x ranged from 1 to 100 ms) extracted from ECG data were taken as fundamental signals. There were obvious differences in distribution of the pRRx serials between 20 people with normal arrhythmia (group I) and 20 patients with arrhythmia (group II). By analyzing the linear indexes and nonlinear indexes of pRRx serials, the computed results show that there are significant statistical differences between the two groups. The linear indexes (AVRR, rMSSD, SDSD) are very different (P<0.001). The nonlinear indexes, from the entropy measures (Sdh, Sph, Spf) and the fractal dimension measures (Dsf, Dcf, Dvm, Drms), also maintain apparent differences (P<0.001). Therefore, the proposed pRRx-serial analysis can characterize the linearity and nonlinearity of the cardiac system to some extent, and can be effective in recognizing the arrhythmia and even heart-related diseases.