Abstract: This paper introduces a car-borne road information collecting and update system (CRICUS).This system is capable to collect road network information and create digital road network effectively by means of integration of GPS and GIS. With a car-borne GPS,the trajectory of the car can be logged while the car runs along the road.5m differential positioning accuracy can be reached by using the cheapest GPS receiver with differential ability.Furthermore,by using carrier phase differential technique,decimeter to centimeter level accuracy can be obtained within a range of several hundred kilometers.In addition,the road-relative information can also be recorded along with the GPS position data to generate road network database. This system consists of two components:field data collecting module and post-mission data processing module. Field data collecting module is composed of a GPS reference station and a mobile unit mounted on a vehicle.The mobile unit includes a GPS receiver and a computer with a field data collecting software installed.The field data collecting software is capable of collecting road spatial and attribute data,storing all data in the hard disk of computer while the car moving,and then transferring all data to computer in the office for subsequent data processing in post-mission data processing module.The GPS data will be recorded automatically and attribute data will be entered through keyboard or voice.The implementation of voice input will make the field work more efficient especially when the car is running on uneven road.The post-mission data processing module has three main functions:GPS data processing,attribute data editing and data output.GPS data will be differentially corrected and smoothed,then the GPS positions will be corrected to the centerline of the road.The attribute editing will retrieve the attribute value which is recorded in voice form and modify the attribute value wrong entered by keyboard in the field.Data output will arrange the spatial and attribute data into a road network database and then export to GeoStar.Preliminary test shows that 1 meter level accuary can be achieved by this system.