In this contribution, we conceive a novel space-time block coded polarization and spatial modulation (STBC-PSM) scheme for multiple-input multiple-output (MIMO) transmissions by reaping the respective advantages of the two techniques. Both quasi-static and fast fading polarized channel models are developed with cross-polar discrimination, cross-polarization coupling, spatial correlations and channel estimation error incorporated into the modeling process to achieve a more accurate characterization of the channel properties. Additionally, a low-complexity maximum likelihood (ML) detector that is able to achieve the optimal error performance is designed for the proposed system. Furthermore, a closed-form upper bound for the bit error rate (BER) is derived, which facilitates the systematic optimization of the polarization patterns by unraveling the critical factors that shape the system performance. Simulation results demonstrate that the proposed STBC-PSM system is capable of achieving a superior BER performance compared to conventional polarization and spatial modulation (PSM) systems and exhibits a robust performance under various channel conditions.