TGF-β1 is produced as a latent cytokine that requires activation to exert its pleiotropic activities. All cells produce latent TGF-β1 but only very few can activate the cytokine via cell type-specific mechanisms. In cancer, TGF-β1 favors disease progression by suppressing anti-tumor T cell responses. Which cells produce immunosuppressive TGF-β1 in human tumors is not known. Putative sources include GARP-expressing cells, which comprise mostly activated Tregs and blood endothelial cells. GARP is a transmembrane protein that forms a covalent complex with latent TGF- β1 (GARP:TGF-β1) on the cell surface and which enables activation of TGF-β1 by integrins. Our laboratory developed monoclonal antibodies that block the activation of TGF-β1 on the surface of GARP-expressing Tregs and, when combined with anti-PD-1 therapy, restore anti-tumor responses in mouse tumor models that were otherwise resistant to anti-PD-1 monotherapy. Phase 1 and 2 trials currently assess the clinical impact of one of these anti-GARP:TGF-β1 antibodies in patients suffering from various solid tumors. However, the prevalence of GARP-expressing Tregs in human tumors is still unclear. While GARP-expressing Tregs have been shown to activate TGF-β1 in mouse models, it remains uncertain whether the same applies to human tumors. Moreover, the role of GARP expressed on blood endothelial cells is poorly understood. We developed an experimental approach combining multiplexed immunofluorescence and computerized image analysis to study the spatial relationships between sources and targets of TGF-β1 signaling in human tissues. We used this approach in the analysis of more than 100 tumor samples from 5 cancer types (colorectal, urothelial, lung and breast primary carcinomas and cutaneous melanoma metastases), compared to patient-matched adjacent non-cancerous tissues and tonsils. We conclude that, in human tumors, GARP-expressing Tregs rather than blood endothelial cells appear to represent a source of TGF-β1 acting on nearby tumor-infiltrating T cells. This might contribute to a form of local immunosuppression that could be reversed with anti-GARP antibodies. Our approach could help distinguish patients with tumors heavily or poorly infiltrated by GARP-expressing Tregs in order to select those that are more likely to respond.