This paper investigates congested traffic loading on long-span bridges through the use of traffic microsimulation. Six months of Weigh-In-Motion (WIM) free-flow traffic data (including cars) are used as input for the microsimulation of congested traffic. Key parameters that affect traffic loading are identified in the output of the microsimulation, and these parameters form the basis for a more computationally-efficient 'pseudo-microsimulation of congested traffic' (PMCT) model. This PMCT model is shown to replicate the traffic loading from full microsimulation accurately and allows longrun simulations, equivalent to 1000 years of congested traffic, to be performed with an acceptably short duration. This reduces the significant uncertainties associated with extrapolating short-run simulation results to long return periods. The 1000-year simulated results from the PMCT are compared with the extrapolated results from full microsimulation, and with the traffic loading from some design codes, for different bridge lengths. Both types of microsimulation are also applied to calculate maximum lifetime loading for two typical long-span bridgesone cable-stayed and one suspension bridgeusing influence lines determined from finite element models.