Abstract: A method for determining the classification thresholds of service levels based on the ordinal Logistic regression model was proposed. The passenger volume estimation method of different station spacing for bus rapid transit(BRT)was proposed， and the cubic function of station spacing and bus running speed was established to calibrate the BRT simulation parameters. The software of VISSIM was used to establish BRT simulation systems with different station spacing of 500~1500m. Based on this， the BRT travel video clips including four stages of arrival， waiting， boarding and departure were produced. The video experiments were carried out and the data of service levels under different station spacings were collected with the sample size of 126. The service level classification of BRT station spacing was established using the proposed classification thresholds method. The thresholds of A~F service levels are 300， 460， 850， 1300 and 1900m respectively. The built service level classification can guide the setting of station spacing under BRT planning and construction， and can be used to evaluate the service levels provided by the current station spacing.

Key words: station spacing; service level classification; ordinal Logistic regression model; classification threshold; passenger volume