Malignant pleural mesothelioma (MPM) comprises three histologic types: epithelioid, sarcomatoid, and biphasic (admixed epithelioid and sarcomatoid). Histologic classification is prognostic, but the underlying biology, particularly determinants of biphasic type, is poorly understood. Previous work using next generation sequencing of bulk tissue characterized the predominant genetic lesions and identified epithelial-mesenchymal transition and inflammation signatures as major axes of molecular variation. However, substantial heterogeneity exists in malignant and immune components of the tumor microenvironment (TME) within any given patient. Thus, improving prognostic accuracy and predicting response to immune and other therapies requires single-cell sampling of multiple well-annotated tumor sites to account for the complex and dynamic TME of MPM. We optimized a tissue workflow to perform in-depth TME characterization of 93 anatomically mapped samples freshly allocated from 36 MPM tumor resections using histopathology coupled with adjacent bulk and single-cell RNA-sequencing. We profiled 19 epithelioid, 13 biphasic, 3 sarcomatoid cases, and 1 MPM tissue without current malignancy and 3 non-MPM pleural specimens as controls. Our workflow generated approximately 2,832 single cell libraries per sample corresponding to 6,495 per case on average (total: 266,265 cells). With this approach, we demonstrate that substantial histopathologic, tumor microenvironment, and clonal (by inferred CNVs) variation exists between multiple samples from a single MPM tumor. We find that previously reported histomolecular gradients are properties of bona fide tumor. In select biphasic cases, we observe that these gradients differ in distinct tumor clones. Future work will assess formally whether any association exists between clonal variation and the tumor-specific epithelial-mesenchymal gradient.