The expected increase in unmanned aerial vehicle (UAV) traffic in European airspace raises concerns regarding the human factors of tower controllers. Dynamic geofences offer tower control a means to safely separate UAVs from manned traffic, without direct interactions with individual UAVs. A preliminary support interface was designed, supporting the operator in separating UAV traffic from manned traffic, while minimising impact on (high priority) UAV efficiency. The effects of traffic conditions and geofence size on control behaviour, safety, efficiency and interface usage were investigated in a human-in-the-loop experiment with active tower and air traffic controllers. Results show that geofences are considered a useful tool in maintaining safety, that larger geofences significantly increase average traffic separation and that the effect on efficiency differs per traffic scenario. Performance could be improved by increasing transparency and predictability of UAV routing, for example by optimising the geofence structure around the runway and by more clearly presenting high-level UAV information to the controller. Further work is needed to investigate controller behaviour and performance in an environment with control over both UAV and manned traffic, considering the temporal aspect of geofences, as well as a broader range of UAV missions and capabilities.