Background: Neuroblastoma is a childhood tumor that is usually fatal after relapse in high-risk patients. Most clinical trials of CAR-T therapy for neuroblastoma are based on targeting the disialoganglioside GD2. B7-H3, a protein from the immunoglobulin superfamily, is a specific marker for neuroblastoma and a number of other solid tumors. We conducted a preclinical study of three variants of anti-B7-H3 CAR-T cells in order to justify the selection of the best candidate for subsequent clinical trials. Methods: The expression level of B7-H3 was measured in a number of cell lines and neuroblastoma tissue samples via flow cytometry. The functional activity of CAR-T cells was compared using an NFAT-inducible reporter assay, a cytotoxicity test, cytokine production, and a repeated stimulation assay. Results: The obtained CAR-T cells carrying all resulting CAR variants specifically recognized and killed B7-H3-positive tumor cells in vitro. Nevertheless, TE9-28z and 8H9-28BBz demonstrated superior activation and cytokine production compared to the second-generation 8H9-BBz construct. TE9-28z and 8H9-28BBz exhibited functional differences in expansion, exhaustion markers, and cytokine secretion in co-cultures with target cells in vitro. In particular, TE9-28z induced higher IFNγ production, while 8H9-28BBz showed increased TNFα release. Despite comparable cytotoxicity, TE9-28z and 8H9-28BBz CAR-T cells exhibited varying persistence depending on the tumor type, and showed signs of functional exhaustion upon prolonged exposure to the target antigen. Conclusion: TE9-28z and 8H9-28BBz were selected for further preclinical development as promising candidates for the effective targeting of B7-H3-expressing malignancies.