Advances in human technology can lead to widespread and rapid increases in interactions between wildlife and potentially disturbing stimuli. The recreational use of drones is widespread and increasing, yet laws and codes of practice which aim to manage deleterious impacts (e.g. negative interactions with wildlife) are reactionary, unscientific and inadequate. One prominent potential negative effect of drones interacting with birds is disturbance; the disruption of normal states caused by responses such as escape. We measure avian escape responses to an approaching drone (n = 561 across 22 species) to inform the development of a code of practice to manage drone‐induced disturbance. Approaches were made at a relatively high and low altitude (10 and 4 m), and at different take‐off distances, both of which are candidate prescriptions for a code of practice. Flight‐initiation distance varied between species, but not between the altitudes tested. The probability of eliciting an escape response was high, and 14.6% higher at the lower altitude (at which 88.4% of overflies resulted in an escape response). Our response rates (from terrestrial and aquatic species) are higher than those reported for different birds in other places, most of which were water or seabirds. The probability of a drone take‐off in itself eliciting a response was low (<20%) when the drone take‐off was >40 m away, and decreased further with increasing distance from birds, with no escapes occurring >120 m. Policy implications. For our sample, no drone take‐off closer than 100 m, and no flight within 100 m would eliminate the vast majority of escape responses by birds. Required separation distances between drones and wildlife may exceed those required for other human activities, such as for walkers.