Wavelet transform coherence (WTC) is a method originally introduced to discover transient linear correlations in random processes. We adjusted and modified the WTC to establish it as a tool for the evaluation of the autonomic nervous system (ANS) activity by applying a bivariate analysis of the cardiorespiratory signals. The WTC provides insight into the transient linear order of the system through the computation of time-frequency maps. Quantitative methods for specific time or frequency bands are also introduced. The WTC was implemented on data from an experimental protocol which included graded exercise stress testing (GXT) with 8 normal subjects. The WTC time-frequency map exhibited a novel presentation of the dynamical changes in RSA phenomena. Furthermore, an integration over the respiratory frequency band revealed a significant (p < 0.001) drop of HR-Respiration coherence around 75% of the maximum effort.