Quantifying respiratory sinus arrhythmia (RSA) can provide an index of parasympathetic function. Fourier spectral analysis, the most widely used approach, estimates the power of the heart rate variability in the frequency band of breathing. However, it neglects the time-varying characteristics of the transitions as well as the nonlinear properties of the cardio-respiratory coupling. Here, we propose a novel approach based on Hilbert-Huang transform, called the multimodal coupling analysis (MMCA) method, to assess cardio-respiratory dynamics by examining the instantaneous nonlinear phase interactions between two interconnected signals (i.e., heart rate and respiration) and compare with the counterparts derived from the wavelet-based method. We used an online database. The corresponding RSA components of the 90-min ECG and respiratory signals of 20 young and 20 elderly healthy subjects were extracted and quantified. A cycle-based analysis and a synchro-squeezed wavelet transform were also introduced to assess the amplitude or phase changes of each respiratory cycle. Our results demonstrated that the diminished mean and standard deviation of the derived dynamical RSA activities can better discriminate between elderly and young subjects. Moreover, the degree of nonlinearity of the cycle-by-cycle RSA waveform derived from the differences between the instantaneous frequency and the mean frequency of each respiratory cycle was significantly decreased in the elderly subjects by the MMCA method. The MMCA method in combination with the cycle-based analysis can potentially be a useful tool to depict the aging changes of the parasympathetic function as well as the waveform nonlinearity of RSA compared to the Fourier-based high-frequency power and the wavelet-based method.