The reliability of CNOSSOS-EU methodology for evaluation of railway noise is dependent on the quality ofinput values that define the system’s excitation and transfer functions. Currently available input values are limited and of insufficient quality, resulting in large errors in noise level calculations. A case study wherein railway rolling noise mitigation measures are evaluated with both CNOSSOS-EU and TWINS show deviations in resulting sound power levels of up to 8 dB(A) and 4 dB(A) for absolute and relative measures respectively. An approach to process TWINS output to extract transfer functions as input values for CNOSSOS-EU is described in this paper. The updated CNOSSOS-EU calculations correspond well with those of TWINS, with the maximum deviation limited to 1.5 dB(A) and 1 dB(A), respectively, for the same absolute and relative measures.