Road section speed of navigation platform and GPS speed of floating car are common sources of road speed data. Using one of the data alone for related research will have problems such as low coverage of real-time update data， unstable data quality and so on. Therefore， a real-time dynamic weight data fusion method based on data reliability is proposed， which fuses the navigation platform road section speed and floating car GPS speed. Before the two kinds of data fusion， space and time are synchronized， and different data reliability evaluation algorithms are designed for the navigation platform road section data and floating car GPS data. Then， the real-time updated navigation platform road section speed and floating car GPS speed are fused based on data reliability to get the final road section speed. The experimental results show that： the average time granularity data updating network coverage rate of navigation platform road section data and GPS road section data are 60.32% and 33.43% respectively， the fusion data based on data reliability is 74.29%， and the average error of road section speed is reduced to 3.82 km/h， which are better than navigation platform road section speed， GPS road section speed and data fusion speed according to fixed weight.