Abstract Introduction: Sub-populations of cancer-associated fibroblasts (CAFs) and perivascular cells in colorectal cancer remain poorly characterized. Combined single cell RNA seq and multiplex antibody staining of human tumors has the potential to identify clinically relevant subsets, where spatial characteristics and outcome associations can provide suggestions regarding function. Results: Four major groups of CAFs were identified based on PDGFRA, FAP and TF; A1 (PDGFRA+/FAP-/TF-), A2PDGFRA-high (PDGFRA-high /FAP+ TF-), A2PDGFRA-low (PDGFRA-low /FAP+ /TF-), A3 (PDGFRA+/FAP-/TF+). CAF subsets showed different spatial associations, such that FAP-negative CAFs were spatially associated with proliferating T-cells and low-proliferating cancer cells, whereas the FAP-positive subsets instead were spatially enriched with non-proliferating T-cells and high-proliferating cancer cells. Furthermore, the FAP-negative cells were enriched around T-cell-infiltrated cancer islands, whereas the FAP-positive subset showed the opposite pattern. Survival analyses showed contrasting compartment-specific good- and bad-prognosis associations of the A1 and A2PDGFRA-low subsets. As for perivascular cells, four cell types defined by M-CAM, MYH11 and PDGFRB were identified. Spatial gradients of cell types, in relationship to endothelial cells, was quantified using novel methods for radial peri-endothelial profiling. MYH11-negative cells showed stronger spatial association with endothelial cells than MYH11-positive cells, and this spatial association was strongest for the PDGFRB-positive subset. Using un-supervised clustering analyses, seven “vascular wall phenotypes” was identified with different radial composition of the four cell subsets. Survival analyses showed that Cluster 4, dominated by the two PDGFRB-positive subsets, was associated with bad prognosis. Summary: Using novel approaches for spatial profiling, this study identifies novel multi-marker subsets of CAFs, and previously unrecognized complex “vessel wall phenotypes”. These show distinct and different associations with cancer cell proliferation, T-cell proliferation, T-cell infiltration into cancer cell islands and survival. Findings prompt continued functional studies on molecular mechanisms underlying these associations, and further exploration of CAF subsets and “vessel wall phenotypes” as predictive markers for immunotherapy and anti-angiogenic drugs. In the future, studies of the spatial transcriptome of different vascular clusters co-localised with immune cells will further enhance the mechanisms of specific cellular interactions.