Assessing railway efficiency is complex for a number of reasons. Railways produce a wide range of outputs including passenger service, freight service and, in some cases, separated infrastructure access services. Railways that differ in scale or in the mix of these services inherently differ in their apparent “efficiency.” Railway data sets, though probably more detailed than in other modes, are fraught with issues of quality, consistency and cost and asset allocation. Assessing “efficiency” necessarily requires both cross-sectional indices to put each railway into proper context and time series data to show changes in performance over time in response to changes in the railway’s economic and policy environment. This paper assembles a wide database of railway data relating to operating scale and various indices of performance over the period of 1970 to 2011. We show, as expected, that railways differ widely in scale and mix of services, which may partly explain differences in ranking by performance indices. We show also that railway performance has changed greatly over time and that, in some cases, changes in performance can at least partly be attributed to reforms in structure, ownership and management incentives.