Diversion model and disaggregate behavioral model of logit type are adopted as aggregate and disaggregate estimates, respectively. Diversion model is assumed as a binary choice process including diversion ratio function at each step in the process. Diversion ratio is assumed as a function of generalized-travel-time ratio. The function is identified with each of the two steps of diversion; diverson from rail and bus to car at first and from the rest undiverted to bus at second. The data used are those on inter-regional occupational person trips. Each function are found enough and/or critically significant in the statistical sense. At the begginning, sixteen characteristics variables are enrolled in disaggregate model, which are identified by use of the above data. Six variables are reached finally and are all reasonable. A brief comparison of goodness of fit to the data are made between two models.