With the increasing use of satellite and ground-based high-spectral-resolution (HSR) measurements for weather and climate applications, accurate and efficient radiative transfer (RT) models have become essential for accurate atmospheric retrievals, for instrument calibration, and to provide benchmark RT solutions. This study develops a spectral data compression (SDCOMP) RT model to simulate HSR radiances in both solar and infrared spectral regions. The SDCOMP approach “compresses” the spectral data in the optical property and radiance domains, utilizing principal component analysis (PCA) twice to alleviate the computational burden. First, an optical-property-based PCA is performed for a given atmospheric scenario (atmospheric, trace gas, and aerosol profiles) to simulate relatively low-spectral-resolution radiances at a small number of representative wavelengths. Second, by using precalculated principal components from an accurate radiance dataset computed for a large number of atmospheric scenarios, a radiance-based PCA is carried out to extend the low-spectral-resolution results to desired HSR results at all wavelengths. This procedure ensures both that individual monochromatic RT calculations are efficiently performed and that the number of such computations is optimized. SDCOMP is approximately three orders of magnitude faster than numerically exact RT calculations. The resulting monochromatic radiance has relative errors less than 0.2% in the solar region and brightness temperature differences less than 0.1 K for over 95% of the cases in the infrared region. The efficiency and accuracy of SDCOMP not only make it useful for analysis of HSR measurements, but also hint at the potential for utilizing this model to perform RT simulations in mesoscale numerical weather and general circulation models.