In this paper, we present an analysis of the impact of perceptions on transport mode choice. Traditional revealed preference surveys very often only involve questions leading to the collection of quantitative data, discrete or continuous, but little attention is given to qualitative data. In the survey on which this research is based, such data appears as adjectives describing a series of transport modes, freely reported by respondents. The calibration of an integrated choice and latent variable model has shown that travelers' perception of comfort in public transports has a significant impact on their mode preferences. Moreover the prediction of individuals' choices by such model is much more accurate than by a logit model with multiple alternatives.