Sound levels from passing trains were recorded. Both maximum A-weighted sound pressure levels L A, max and equivalent levels L Aeq over 60 second time periods containing each pass-by were measured. Recordings of 15 pass-bys were made at each of two different sites. At each site attenuation over level, grass-covered ground and through shelter belts was measured 1.5 m above ground. Attenuation differences at each site were due to both minor variations in terrain configuration (track above/below adjacent terrain) and to attenuation in vegetation. The latter cannot be separated. L Aeq attenuations were smaller than L A,max attenuations, as should be expected. Noise reduction by shelter belts—i.e., the approximate difference between attenuation over grass-covered ground and thouugh vegetation, respectively—was nearly the same, expressed in L A, max and L Aeq values. Shelter belts selected for this investigation were carefully maintained. Their overall structure, therefore, was very uniform. Behind a dense, 15 year old shelter belt, 50 m wide, consisting of beeches and various conifers planted between older birches and elms, noise levels were 8 to 9 dB lower than in level grass-covered country. Behind a dense, 10 to 20 year old shelter belt, 25 m wide, consisting of oaks, hornbeams, poplars, silver firs and various sorts of bushes, noise levels were 6 or 7 dB lower than in level grass-covered country. The attenuations measured seemed to be of such an order of magnitude that similar belts of trees and bushes could be a means of practical noise reduction. Further investigations, therefore, seem to be worthwhile.