Over the years microscopic traffic simulation has evolved as the premier tool to analyze complex and congested transportation networks. However, despite the robustness and wide spread use of traffic microsimulation, some gaps and limitations still exist that can affect the accuracy of these models' results. Moreover the change of traffic characteristics and driver behavior during the transition from undersaturated to saturated condition is not completely understood. This dissertation addresses these two issues. The road network chosen for the microscopic simulation and field data collection is a six lane main traffic artery located in Tuscaloosa, Alabama, USA. The entire research work contains three related research efforts, each conducted along the topic of this dissertation. The first research thrust focused on the sensitivity and accuracy of the microscopic traffic simulation. Specifically it investigated the sensitivity of MOEs to simulation initialization time, required number of repetitions, and major contributors of variation in MOEs. The second research thrust dealt with field investigation of operational parameters including gap acceptance and lane changing during different levels of traffic flow. The final research effort explored the variations in simulation results using existing embedded/default values of lane change parameters (lane change duration and look ahead distance), versus using values obtained from field observation for both free flow and saturated traffic conditions. From all the research efforts, the following broad conclusions were drawn, * Traffic flows at signals that are approaching saturation are still complex to analyze, and the interactions between traffic parameter are not well understood. * When traffic flow on a typical arterial approaches saturation, drivers take higher risks (eg: drivers accept smaller gaps). * A statistical analysis of gap acceptance and lane changing confirmed what is suspected intuitively. * Existing traffic microsimulation tools simplify some of the traffic parameters in simulation models. These parameters may be recoded or recalibrated for better accuracy of simulation results. * In traffic microsimulation an increased number of simulation runs certainly helps in stabilizing the variability of the MOE and it is advisable to use a longer simulation time (eg. 60 minutes) to reduce the variation of MOEs.