In parts of Australia where the clearing of native vegetation has been widespread the establishment of windbreaks is a common farming practice. There is some evidence that birds respond to certain characteristics of windbreaks, particularly width. This is the first study to investigate such effects in Australia. Six properties, containing 61 windbreak sites, were surveyed for birds 12 times between 1993 and 1997. Generalised linear models were used to test relationships between 19 predictor variables and bird community attributes, including the densities of 14 species of birds. Canonical variate analysis was undertaken to discriminate among four types of windbreaks and their avifauna. Species known to be declining in the region and uncommon (<100 observations) woodland birds were more abundant in wider and more floristically diverse windbreak sites. Several native species had a positive response to tree-height, density of understorey, structural diversity, the proportion of native plant species in the site or the proximity of remnant bushland. The density of introduced bird species, as a group, was strongly negatively related to the proportion of native plants and, for one species, Common Starling (Sturnus vulgaris), to width of windbreak. Recommendations for windbreak design are provided and include: using a diversity of native plants; providing adequate width, preferably over 20 m; and having a diversity of understorey densities. It is also recommended that patterns of experimental windbreaks are created that could be used for formal hypothesis testing.