We present a general multistep procedure for the determination of wavelength/wavenumber-dependent sensitivity of Raman spectrometers equipped with multichannel detectors. In this approach, we start with photons-per-wavenumber conversion (C0) followed by a correction (C1) for the channel-to-channel variation in the sensitivity. The final correction (C2) is determined from a set of observed intensity ratios for standard Raman bands of selected molecules, compared with the analogous reference set of their accurate theoretical counterparts. The developed approach is employed here for calibrating our Raman spectrometer in the high-wavenumber region using the vibration–rotation Raman bands of H2, HD, and D2. Theoretical intensity ratios for these bands were computed using accurate ro–vibrational matrix elements of polarizability invariants. The presented scheme allowed us to significantly improve the accuracy of the relative Raman intensities for measured samples. The accuracy was examined by determining the Boltzmann temperatures from the Raman intensities and comparing them with the corresponding thermocouple measurements. The uncertainty of such a Raman thermometer was found to be within 5%, confirming the validity and robustness of the proposed intensity calibration scheme. The presented intensity correction goes beyond the scope of common calibration schemes, traditionally limited to emission standards like tungsten lamps. While the first two steps of our calibration process (C0 and C1) are fairly typical, the additional correction (C2) derived from Raman spectral intensities seems to constitute an important novel step towards the standardization of Raman spectroscopy.