Understanding the temporal patterns in groundwater levels and their spatial distributions is essential for quantifying the natural and anthropogenic impacts on groundwater resources for better management and planning decisions. The two most popular clustering analysis methods in the literature, hierarchical clustering analysis and self-organizing maps, were used in this study to investigate the temporal patterns of groundwater levels from a dataset with 910 observation bores in the largest river system in Australia. Results showed the following: (1) Six dominant cluster patterns were found that could explain the temporal groundwater trends in the Murray–Darling Basin. Interpretation of each of these patterns indicated how groundwater in each cluster behaved before, during, and after the Millennium Drought. (2) The two methods produced similar results, indicating the robustness of the six dominant patterns that were identified. (3) The Millennium Drought, from 1997 to 2009, had a clear impact on groundwater level temporal variability and trends. An example causal attribution analysis based on the clustering results (using a neural network model to represent groundwater level dynamics) is introduced and will be expanded in future work to identify drivers of temporal and spatial changes in groundwater level for each of the dominant patterns, leading to possibilities for better water resource understanding and management.