<p>Understanding the links between large-scale climate factors and regional meteorological drought is essential for increasing the accuracy of drought prediction and implementing effective prevention strategies. However, methods based on nonlinear and multivariate causality analysis have not yet been fully explored. This study investigated the causal effects of ten large-scale climate indices on meteorological drought in the Pearl River Basin (PRB) from 1961 to 2022 via multivariate partial mutual information from mixed embedding (PMIME). The standardized precipitation evapotranspiration index (SPEI) was used to represent meteorological drought in the PRB, while empirical orthogonal teleconnections (EOTs) were employed to extract SPEI patterns. After comparing the results of mutual information (MI) analysis, bivariate causal analysis via PMIME, and multivariate causal analysis via PMIME, we identified two key findings. First, EOT analysis revealed five significant SPEI patterns in the PRB, accounting for 48.84%, 19.57%, 13.22%, 4.47%, and 2.53% of the total variance. Not all of the global climate indices studied exhibited a causal effect on each EOT. Second, owing to the interactions among variables, the results of multivariate PMIME differ from those of MI analysis and bivariate causal analysis via PMIME. The results of multivariate PMIME indicated complex interactions among the studied global climate indices for EOTs 1, 2, 4, and 5. For EOTs 1, 2, 3, 4, and 5, the global climate indices with the strongest causal effects were the Arctic Oscillation (AO), AO, West Pacific Pattern, Nino 3.4 sea surface temperature index, and South Indian Ocean Dipole Index, respectively. These findings are highly important for understanding the teleconnection between large-scale climate patterns and meteorological drought in the PRB.</p>