This paper presents attempts to improve some defects of conventional disaggregate models by using attitudinal data for travel choice prediction. Our first approach is the use of attitudinal data as explanatory variables for qualitative factor. The second approach is it's use as an index for segmentation to consider variation of individual decision criteria. Our conclusions are as follows. As explanatory variables evaluated values and reported values of LOS variables have good explanatory power and evaluated values, in particular, have ability to explain qualitative factor. The segmentation based on whether person evaluates highly the delay of travel time or not is effective in explaining variation of individual choice.