The multiwavelet scale multidimensional recurrence quantification analysis (MWMRQA) method is proposed in this paper, which is a combination of multidimensional recurrence quantification analysis and wavelet packet decomposition. It allows us to quantify the recurrence properties of a single multidimensional time series under different wavelet scales. We apply the MWMRQA method to the Lorenz system and the Chinese stock market, respectively, and show the feasibility of this method as well as the dynamic variation of the Lorenz system and the Chinese stock market under different wavelet scales. This provides another perspective for other disciplines that need to study the recurrence properties of different scales in the future.