The use of large animal spontaneous models of solid cancers, such as dogs with osteosarcoma (OS), can help develop new cancer immunotherapy approaches, including chimeric antigen receptor (CAR) T cells. The goal of the present study was to generate canine CAR T cells targeting the B7-H3 (CD276) co-stimulatory molecule overexpressed by several solid cancers, including OS in both humans and dogs, and to assess their ability to recognize B7-H3 expressed by canine OS cell lines or by canine tumors in xenograft models. A second objective was to determine whether a novel dual CAR that expressed a chemokine receptor together with the B7-H3 CAR improved the activity of the canine CAR T cells. Therefore, in the studies reported here we examined B7-H3 expression by canine OS tumors, evaluated target engagement by canine B7-H3 CAR T cells in vitro, and compared the relative effectiveness of B7-H3 CAR T cells versus B7-H3-CXCR2 dual CAR T cells in canine xenograft models. We found that most canine OS tumors expressed B7-H3; whereas, levels were undetectable on normal dog tissues. Both B7-H3 CAR T cells demonstrated activation and OS-specific target killing in vitro, but there was significantly greater cytokine production by B7-H3-CXCR2 CAR T cells. In canine OS xenograft models, little anti-tumor activity was generated by B7-H3 CAR T cells; whereas, B7-H3-CXCR2 CAR T cells significantly inhibited tumor growth, inducing complete tumor elimination in most treated mice. These findings indicated therefore that addition of a chemokine receptor could significantly improve the anti-tumor activity of canine B7-H3 CAR T cells, and that evaluation of this new dual CAR construct in dogs with primary or metastatic OS is warranted since such studies could provide a critical and realistic validation of the chemokine receptor concept.