Abstract The literature on multiple windbreaks is reviewed in an attempt to explain observed differences in wind protection achieved by two different orchard management systems. The level of wind-rub blemishes on kiwifruit is used as an indication of the accumulated mechanical stress on the vine. Within a traditionally sheltered and trellised orchard, wind damage increased rapidly with distance downwind from close-spaced artificial shelter. Badly damaged vines had up to 58% of their export-sized fruit rejected because of wind-rub, the average loss being 15%. In striking contrast, vines trained on a nearby Tatura trellis with minimal use of windbreaks lost only 1% of the total crop because of wind-induced blemishes. Wind-rub damage is caused by oscillatory motions of fruit forced by turbulent wind fluctuations. Fruit behave as underdamped systems with damping coefficients near 0.03 and resonant frequencies close to 2 Hz. Calculations of the reduction in turbulent energy at inertial sub-range frequencies using Kolmogorov's formula and velocity profile data from the literature suggest that natural shelter of the type often used in New Zealand should be very effective. This is contrary to our field evidence and we are led to the view that the advection of turbulence in the wake flow, generated by the windbreaks themselves, is responsible for the observed patterns of damage in the traditionally sheltered orchard. On the other hand, the success of the Tatura trellis in conjunction with minimal shelter is attributed to the development of ‘skimming flow’ and the mutual sheltering of adjacent close-spaced rows of foliage.