In order to forecast the traffic demand in a new town,by using the random utility theory,the distributionof residential and activity locations is defined as a Logit formulation of land rent and traffic location,and a balanceequation is established to simulate the assignment of houses and activities.Thus,a Logit model is constructed tosimulate the spatial distribution of land use in the new town.On this basis,a combination model of trip generationand distribution is constructed by embedding the population and employment distribution and the location functioninto the traffic demand forecasting model.Finally,Guangzhou Nansha new town is taken as an example to test andverify the two models.The results show that (1) the Logit model is able to simulate the spatial distribution of landuse and analyze the distribution of the selected residential and activity locations in the new town; and (2) when thevacancy rate of housing and shopping is taken into account,the total population and daily trip generations of Nanshanew town in 2030 are predicted to be 2420 thousand and 7322 thousand respectively,which decrease respectivelyby 19.3% and 12.8% in comparison with the results of the conventional method.