Wakes behind 2D fences and 3D obstacles are reviewed with special emphasis on reduced mean wind speeds and sheltering effects. Based partly on Perera's study of wakes behind 2D fences, and assuming a Gaussian spread for wakes behind 3D obstacles, a shelter model is proposed and tested. The shelter produced depends on a wake moment coefficient C̃k which appears to be significantly less for 3D obstacles than for 2D fences. The model provides a simple basis on which to “correct” anemometer data for sheltering effects associated with upstream obstacles. Such corrections are an important step in the generation of improved surface wind climatologies and wind atlases.