Spatially charting molecular cell types at single-cell resolution across the three-dimensional (3D) volume of the brain is critical for illustrating the molecular basis of the brain anatomy and functions. Single-cell RNA sequencing (scRNA-seq) has profiled molecular cell types in the mouse brain, but cannot capture their spatial organization. Here, we employed an in situ sequencing technique, STARmap PLUS, to map 1.09 million high-quality cells across the whole adult mouse brain and the spinal cord, profiling 1,022 genes at subcellular resolution with a voxel size of 194 X 194 X 345 nm3 in 3D. We developed computational pipelines to segment, cluster, and annotate 230 molecular cell types by single-cell gene expression and 106 molecular tissue regions by spatial niche gene expression. Joint analyses of molecular cell types and molecular tissue regions enabled a systematic molecular spatial cell type nomenclature and allowed the identification of tissue architectures previously undefined in established brain anatomy. To create a transcriptome-wide spatial atlas, we further integrated the STARmap PLUS measurements with a published scRNA-seq atlas, imputing 11,844 genes at the single-cell level. Finally, we delineated the tropisms of a brain-wide transgene delivery tool, AAV-PHP.eB5,6, across the molecular cell types and tissue regions of the whole mouse brain. Together, this annotated dataset provides a comprehensive, high-resolution, single-cell resource that integrates a spatial molecular atlas, cell taxonomy, brain anatomy, and genetic manipulation accessibility of the mammalian central nervous system (CNS).