Context. The orientation and rotation of Mars, which can be described by a set of Euler angles (longitude, obliquity, and rotation angles), is estimated from radioscience data (tracking of orbiters and landers) and is then used to infer Mars internal properties. The data are analyzed using a modeling expressed within the Barycentric Celestial Reference System (BCRS). This modeling includes several relativistic contributions that need to be taken properly into account to avoid a misinterpretation of the data. Aims. We provide new and more accurate (to the 0.1 mas level) estimations of the relativistic corrections to be included in the BCRS model of the orientation and rotation of Mars. Methods. There are two types of relativistic contributions in Mars rotation and orientation: (i) those that directly impact the Euler angles and (ii) those resulting from the time transformation between a local Mars reference frame and BCRS. The former correspond essentially to the geodetic effect, but also to the smaller Lense-Thirring and Thomas precession effects. We compute them assuming that Mars evolves on a Keplerian orbit. As for the latter, we compute the effect of the time transformation and compare the rotation angle corrections obtained assuming that the planets evolve on Keplerian orbits with that obtained with realistic orbits as described by ephemerides. Results. The relativistic correction in longitude comes mainly from the geodetic effect and results in the geodetic precession (6.754 mas yr −1 ) and the geodetic annual nutation (0.565 mas amplitude). For the rotation angle, the correction is dominated by the effect of the time transformation. The main annual, semi-annual, and ter-annual terms have amplitudes of 166.954 mas, 7.783 mas, and 0.544 mas, respectively. The amplitude of the annual term differs by about 9 mas from the estimate usually considered by the community. We identify new terms at the Mars-Jupiter and Mars-Saturn synodic periods (0.567 mas and 0.102 mas amplitude) that are relevant considering the current level of uncertainty of the measurements, as well as a contribution to the rotation rate (7.3088 mas day −1 ). There is no significant correction that applies to the obliquity.