Driving simulators have become an established tool in driver behaviour research by offering a controllable, safe and cost-effective alternative to real world driving. A challenge for using driving simulators as a research tool has been to elicit driving behaviour that equals real world driving. With respect to driver headways few studies have made a direct comparison between behaviour in real and virtual environments. The present study compared driver headway choice in a driving simulator and in an instrumented vehicle. Twenty-two participants carried out instructions to either change their headway to a specific value or to choose a headway as they would normally do. The speed of the lead vehicle (80, 100 or 120 km/h) as well as the target headway (1, 1.5, 2 s) were varied between trials. Specific headway instructions were provided in seconds as well as metres. The attained headways were compared between the virtual and the real environment. Results show no significant difference between headway choice in the simulator and on a real road, neither for self-chosen nor for instructed headways. The results provide support for the use of driving simulators in studies on headway choice. © 2014 Elsevier Ltd. All rights reserved.