Current concerns surrounding regional air pollution, climate change, rising gasoline prices, and urban congestion could presage a substantial increase in the bicycle mode share. However, state-of-the-art methods for the safe and efficient design of bicycle facilities are based on difficult-to-collect data and potentially dubious assumptions regarding cyclist behavior. Simulation models offer a way forward, but existing bicycling models in the academic literature have not been validated with actual data. These shortcomings are addressed by obtaining real-world bicycle data and implementing a multilane, inhomogeneous cellular automaton simulation model that can reproduce observations. The existing literature is reviewed to inform the data collection and model development. It is found that the model emulates field conditions while possibly underpredicting bike path capacity. Since the simulation model can "observe" individual cyclists, it is ideally suited to determine level of service based on difficult-to-observe cycling events such as passing. Future work on data collection and model development is suggested.