Variability in passenger service dwell time at bus stops affects berth occupancy, stop capacity, vehicle queue formation, and adjacent traffic conditions, particularly at stops shared by urban buses, suburban buses, taxis, and other passenger-serving vehicles. Existing approaches generally represent dwell time at an aggregated bus level, while a methodological gap remains in the probabilistic parameterization of its individual components for different user groups sharing the same stop. This study develops and verifies a group-specific probabilistic parameterization of passenger service time components under mixed service conditions. The novelty lies in combining group-specific probabilistic models of passenger demand and post-service technical time with an event-based representation of boarding and alighting within a unified simulation framework. The analysis used 5,953 observations of boarding and alighting time collected at bus stops in Kraków and Kielce, Poland. Boarding and alighting passenger counts, boarding and alighting time, and post-service technical time were analyzed separately by user group. Passenger counts were modeled using discrete probability distributions, while post-service technical time was modeled using continuous distributions. Parameters were estimated using maximum likelihood estimation and maximum goodness-of-fit estimation based on the Anderson–Darling statistic, with uncertainty assessed by bootstrap resampling. For urban buses, the negative binomial distribution provided the best fit for boarding passenger counts at 15 of 18 stops and for alighting passenger counts at 14 of 18 stops. For suburban buses, acceptable fits were obtained at four stops for both boarding and alighting. The negative binomial distribution was adopted as a practical passenger-count model because it frequently occurred among the best-fitting distributions and its parameters could be derived from the mean and standard deviation. In 69 of 78 complete fits, parameters calculated from these statistics fell within the uncertainty range of the maximum likelihood estimates. Post-service technical time was represented by a gamma distribution for urban buses and by a logistic approximation for other users. Simulation-based verification yielded mean absolute errors of 0.81 s for urban buses, 1.91 s for suburban buses, and 2.04 s for taxis and other vehicles. The results support separate parameterization of passenger service time for individual user groups at shared-use bus stops.