Abstract
Increasing use is being made of models which express how the number of trips between a pair of places by a given mode depends on the time or cost of travel by the various available modes. These models have run into various difficulties, often associated with the evaluation of the benefits resulting from changes. A general theory from which a variety of models can be derived by making different assumptions about the statistical distribution of parameters which vary from one person to another and which give rise to observable differences in trip making behaviour, is set out.