Motorcycles constitute a significant proportion of traffic in many countries but are poorly represented in existing traffic flow theories and simulation software. A new approach to modeling mixed traffic is introduced focusing on depicting the movements of motorcycles. In this study, the characteristic patterns of motorcycle behavior were identified, and the key elements contributing to these patterns were extracted. Then three mathematical models were developed to depict these key elements, which were calibrated by using field data collected at Victoria Embankment in central London. After the calibration procedures, these models were integrated into an agent-based simulation model system. The ability of the simulator to reproduce plausible patterns of car and motorcycle behavior was verified. A number of potential applications of this simulator for the management of mixed traffic streams in urban areas are discussed.