This paper reviews evidence from British experience of the valuation of rail crowding obtained over 20 years from 17 studies. It summarizes these studies, places some useful empirical evidence in the public domain and draws lessons from this considerable body of evidence and experience. Crowding valuations, both for standing and seated in crowding conditions, are summarized in terms of time multipliers, which are inherently more transferable than monetary equivalents. A meta-analysis of 208 valuations is reported, finding the valuations to vary with load factor and journey purpose. The seating multiplier averages 1.19 and the standing multiplier averages 2.32. The latter is in line with widely used multipliers applied to walking and waiting time. The most recent evidence is based around the number of standing passengers per square metre, thereby providing a more accurate measure of the discomfort of standing since, unlike load factor, it allows for the layout of the carriage and ease with which crowding can be accommodated. As far as methodology is concerned, the paper covers issues such as presenting crowding in 'stated preference' exercises and the realism of the crowding levels offered, non-linearities in the relationship between crowding multipliers and the severity and amount of crowding time, and probabilistic versus deterministic representations of crowding. The paper also identifies future areas for research.