In this study, the manner of private taxis drivers has been investigated for choosing passenger and destination from a fixed point. Therefore, two models called Multinomial and Nested Logit Models have been utilized. The information gained by scrolling in 2016 is the input data, which are in the format of revealed preference, acquired by the verbal interview in Vanak Square in Tehran (Iran). Based on data resulted by 120 questionnaires the selection modeling of passenger and destination was done. The results of the descriptive analysis show that 96.7% of respondents are men and only 3.3% are women. In addition, 15% of them are illiterate, 16.7% have under Diploma degree, 52.5% have Diploma degree and 15.8% have Bachelor degree or higher. On average, the verification was 44%, and the results achieved by this research show that the parameters consisting of searching time, the working time of drivers and the traveling time were the most important factors resulted from the calibration of the Logit models. Nested Logit model has a better performance compared with the Multinomial Logit model. The Nested Logit Model has ρ2=0.45 while this value is 0.35 for the Multinomial Logit Model. Finally, the suitable decision has been made in the various path based on gained results.