A numerical comparison is made between the road traffic noise shielding provided by 4-m high vertically erected walls and (earth) berms, averaged over large zones behind them. A previously developed and validated full-wave numerical sound propagation model was used. In absence of wind, a noise screen is preferred when it can be placed at the same position as the foot of the berm at the source side. In case of a fixed top position for both the wall and berm, an acoustically soft berm with a flat top gives similar shielding as the wall. In case of downwind sound propagation (i.e. a worst-case situation), the noise wall efficiency largely decreases. Strong wind might lead to an almost complete loss of noise wall shielding when compared to sound propagation over unobstructed terrain in absence of wind. In contrast, with decreasing berm slope angle, downward refraction by wind can become very small. In case of berms with a slope of 1:3, or berms with steeper slopes but with a flat top, the averaged wind effect can be smaller than 1 dBA because of the limited magnitude of vertical gradients in the horizontal component of the wind speed near such more streamlined obstacles. When looking at long-term equivalent noise levels, including periods with wind, acoustically soft and shallow berms should be chosen upon vertically erected noise walls. In addition, there are clear non-acoustical benefits associated with the use of such natural berms. Wind-breaking vegetation has not been considered in this study.