Transport analysis and choice modelling have had a fruitful symbiosis over several decades.  Transport planning has needed quantitative forecasts of traffic flows, while the appraisal of transport projects has also required measures of willingness to pay: for both of these objectives appeal has been made to choice modelling.  On the other hand, developments in choice modelling have offered new capabilities for transport analysis.  The paper discusses this interaction, noting the major developments and indicating the current research areas.

A key feature of transport analysis has been the use of revealed preference data, in both disaggregate and aggregate forms.  RP data is often collected for other purposes than modelling, making it a cheap if not always ideal resource, while its existence also raises the expectation that models will be consistent with it.  Combination with Stated Preferences in various forms is fruitful, but raises issues that have not yet been resolved, while issues also remain with RP data itself: these issues are discussed.

 Transport infrastructure is often intended to have a life of many years and to perform well over that period meaning that forecasts of traffic over a lengthy period are needed.  The need for forecasts requires a range of choice modelling techniques to be developed, distinct from the estimation techniques that fill the majority of the literature.  Again, the key points of this work are set out, new developments presented and current research issues outlined.