Abstract Background: Adoptive cell therapies such as CAR T have been curative for many liquid tumors, yet this success has struggled to translate to solid tumors due to the challenges posed by the solid tumor microenvironment (TME). Pooled CAR screens (e.g. CAR-POOLING) have recently emerged as a powerful approach for testing many designs for improving CAR T cell performance. However, these screens are typically restricted to univariate readouts (e.g. abundance) and cannot measure the cell-cell and cell-TME interactions that are critical for CAR T success in solid tumors. Here, we present a platform for performing pooled screens in vivo using spatial biology, facilitating simultaneous measurement of cell autonomous phenotypes, cell-cell interactions, and cell-environment interactions for hundreds of distinct cell therapy designs in a single mouse. Methods: We developed a custom spatial biology workflow to determine the identity of specific CAR constructs, armorings, or combinations in specific individual cells from a microscope image (“design deconvolution”). We combined this design deconvolution with simultaneous multiplexed spatial biology phenotyping in the same cells & samples, followed by computer vision interpretation to quantify each design’s performance across functional categories important for efficacy in solid tumors such as: tumor infiltration, tumor cell killing, proliferation, response to antigen heterogeneity & escape, exhaustion, persistence, and activity in immunosuppressive tumor microenvironments. In multiple mouse cancer models, we quantified the performance of hundreds of different cell therapy designs. Results: Using our pooled screens, we identified designs with superior performance in solid tumors across one or more key functional categories (e.g. infiltration, activity in immunosuppressive TMEs, etc.). We also mapped the landscape of cell therapy design principles by comparing dozens of existing designs derived from the literature head-to-head in the same animal, and we compared these benchmarked designs to novel cell therapies designed to improve CAR T function. We used these data to identify common rules that govern the design-phenotype relationship for CAR Ts in solid tumors and to identify CAR T designs with desired properties in solid tumors. Conclusions: Using a pooled in vivo screening platform, we discovered cell therapies that outperform existing designs across a variety of clinically-relevant metrics. Our spatial multi-omics measurements are critical for identifying designs that can effectively traffic to, infiltrate, and function in the solid tumor microenvironment. We are now using our pooled in vivo screens to identify highly efficacious designs in multiple solid tumor models. Citation Format: Cassandra Kontur, Gundula Povysil, Megha Sah, Mariya London, Nan Chen, Annie Dyatel, Stephen Pu, David V. Phizicky, Xinchen Wang. Pooled in vivo screening of hundreds of T cell therapy designs using spatial multi-omics identifies novel designs with superior TME performance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6329.