Liubao tea and ripen Pu-erh tea are representatives of dark tea in southern China. The two dark teas are famous for unique flavors, but confusingly different in development status of tea industry. In this study, microbial DNA amplification sequencing and mass spectrometry-based untargeted metabolomics were applied to observe significant differences in microbial community structure and metabolite profile between the two teas. The Shannon indices of fungi and metabolites in Liubao tea are higher than those in ripen Pu-erh tea. The dominant bacterial and fungal genera, as well as microbial biomarkers of Liubao tea and ripen Pu-erh tea were identified. The combined statistical and molecular networking analysis shows flavan-3-ols as the discriminating features between metabolite profiles of the two dark teas in level of metabolite family. More importantly, the α and β diversity analysis reveals higher pairwise Shannon index differences and Canberra distances of both microbes and metabolites in Liubao tea than those in ripen Pu-erh tea, indicating greater heterogeneity, or lower quality stability of Liubao tea products. These findings illustrate way to improve protocols of Liubao tea processing, and show urgency of involving molecular networking in workflow of metabolomics research.