The rolling noise generated by railway vehicles was evaluated in scale-model tests. The similarity relations associated with wheel/rail noise were derived and the full-scale phenomena were estimated from the scale-model tests. When converting scale-model findings to full-scale, the frequency is scaled by the ratio 1/n, accelerances of rail and wheel vibration are scaled by the ratio 1/n 3 , and a logarithmic term is added to the sound pressure level. To verify the validity of these scaling relations, the scale-model measurements were compared with the results of field tests. Through measurements with an impact hammer, the vibration characteristics of the track and wheel were estimated in scale-model tests, were scaled using the proposed similarity laws, and were compared with field measurements of actual tracks and wheels. The quantitative trends of the actual characteristics were estimated well from the scale-model results. The field tests showed that, though the measured noise in the test rig had an unsuitable signal-noise ratio because of the greater driven noise of the test rig itself, the rail vibrations measured in the scale-model test did have a suitable signal-noise ratio. The rail vibration measured in the scale-model test is in good agreement with the measurements from the field tests. If noise from the test rig can be sufficiently reduced, or if rolling noise in the scale model is sufficiently large, we conclude that scale-model tests can simulate actual rolling noise.