A variety of short-term traffic flow forecasting methods have been developed in the last two decades. However, due to the complexity of traffic conditions and the drawbacks existing in those methods, predictions of traffic flow generally lack accuracy and robustness. In response to these problems, a fuzzy logic system methodology is proposed in this study. The fuzzy logic system methods are used to forecast traffic flow and compare with existing methods, using dual-loop data collected from I-35 in San Antonio, Texas. Forecasting results show that the fuzzy logic system produces more accurate and stable predictions. It is also more robust as it is able to forecast flow under various traffic and detector operation conditions.