The objective of this paper is to inform the debate on how efficiency targets for Network Rail (formerly Railtrack) should be set during the 2002/03 Interim Review and beyond. Given the problems experienced during the 2000 Periodic Review, which focused on external benchmarks, we propose an internal benchmarking approach, drawing on data for seven geographical Zones within Railtrack (over the period 1995/96 to 2001/02). Our approach mirrors the yardstick competition method used in other UK regulated industries. Three efficiency measurement techniques are applied to this data (DEA; COLS; SFA). Our results suggest that Railtrack (as a whole) delivered substantial improvements in productivity in the early years after privatisation, although these savings were largely offset by the post-Hatfield cost increases. However, looking forward, Zonal efficiency differences suggest that the company could make significant savings in future years by applying (its own) best practice consistently across the network.