The dysfunction of astrocytes in response to environmental factors contributes to many neurological diseases by impacting neuroinflammation responses, glutamate and ion homeostasis, and cholesterol and sphingolipid metabolism, which calls for comprehensive and high-resolution analysis. However, single-cell transcriptome analyses of astrocytes have been hampered by the sparseness of human brain specimens. Here, we demonstrate how large-scale integration of multi-omics data, including single-cell and spatial transcriptomic and proteomic data, overcomes these limitations. We created a single-cell transcriptomic dataset of human brains by integration, consensus annotation, and analyzing 302 publicly available single-cell RNA-sequencing (scRNA-seq) datasets, highlighting the power to resolve previously unidentifiable astrocyte subpopulations. The resulting dataset includes nearly one million cells that span a wide variety of diseases, including Alzheimer’s disease (AD), Parkinson’s disease (PD), Huntington’s disease (HD), multiple sclerosis (MS), epilepsy (Epi), and chronic traumatic encephalopathy (CTE). We profiled the astrocytes at three levels, subtype compositions, regulatory modules, and cell–cell communications, and comprehensively depicted the heterogeneity of pathological astrocytes. We constructed seven transcriptomic modules that are involved in the onset and progress of disease development, such as the M2 ECM and M4 stress modules. We validated that the M2 ECM module could furnish potential markers for AD early diagnosis at both the transcriptome and protein levels. In order to accomplish a high-resolution, local identification of astrocyte subtypes, we also carried out a spatial transcriptome analysis of mouse brains using the integrated dataset as a reference. We found that astrocyte subtypes are regionally heterogeneous. We identified dynamic cell–cell interactions in different disorders and found that astrocytes participate in key signaling pathways, such as NRG3-ERBB4, in epilepsy. Our work supports the utility of large-scale integration of single-cell transcriptomic data, which offers new insights into underlying multiple CNS disease mechanisms where astrocytes are involved.