A multiple sub-pupil ultra-spectral imager system for simultaneous detection of multiple substances is proposed. By employing pupil separation prisms and grating multilevel spectra, the system achieves simultaneous detection of three spectral channels with a single spectrometer and detector, featuring an ultra-high spectral resolution of 0.1 nm. However, due to simultaneous detection of three channels, the system suffers from significant stray light issues. In response to this problem, a UV multiple sub-pupil ultra-spectral imager system optical-machinery model is constructed in this article. The main sources of stray light are determined through theoretical derivation and simulation analysis, and the level of stray light in the system is analyzed using simulation analysis software. A structure for stray light suppression is optimized, and the use of multispectral filters is proposed to suppress stray light generated by spectral crosstalk in the system. The effectiveness of stray light suppression is evaluated based on the energy response of the receiving surface. The analysis results show that after optimization, the level of stray light in the T1 channel is reduced from 1.4% to 0.65‰, in the T2 channel from 3.0‰ to 0.5‰, and in the T3 channel from 0.16‰ to 0.05‰. Therefore, it can be concluded that the proposed method for stray light suppression in the UV multiple sub-pupil ultra-spectral imager system addressed in this article meets the indicator requirements.