Virtual wildlife fencing presents as a cost-effective measure for roadkill mitigation, which aids in reducing fragmentation of wildlife populations by facilitating safer movement of wildlife across the landscape. In this study, we conducted an audit of a virtual fence installation in south-east Queensland, Australia. We assessed its reliability in flows of traffic and the effect that installation parameters and site conditions had on its effective operation in an urban setting. We made observations on the behavioural response of Eastern Grey Kangaroos Macropus giganteus to the acoustic signals produced by the fence. We found that the fencing activated consistently in response to headlights at dusk and dawn, and when traffic flows were dense, despite considerable variations in the range of installation parameters. However, we identified that the response of the virtual fence to headlights was affected by road curvature and we identified inconsistencies in the timing and pattern of activation in response to traffic. Behavioural observations showed a significant increase in kangaroo vigilance in response to the acoustic signal of the fence when resting or grazing, and kangaroos detected the acoustic signal up to 50 m away. While virtual fencing operates effectively and is a low-cost roadkill mitigation option that can be applied to the urban environment, more research is needed to better understand the effect of its acoustic and visual signals on wildlife behaviour and efficacy in busy urban environments.