Abstract: The accuracy and quality of spatial data are one of the fundamental issues in geographic information system(GIS) and geographic information system for transportation(GIS-T).Since the manual digitization is still one of the major methods for capturing the spatial data,error-processing models for the road curves from manual digitization are discussed in this paper.The road curves as the fundamental forms of the road spatial data are composed of regular geometric curves with vigorous mathematics expressions,therefore,it is pertinent to use parametric adjustment approaches to establish error processing models for the digitized road curves.By using least square theory,the parametric adjustment models including error equations and the calculations of parameter approximate values are first derived for the basic geometric features-straight lines,circular curves and transition curves that constitute the road curve alignments.The combined adjustment models based on those of basic geometric features are then presented for the road curve obtaining from manual map digitization.Therefore,the system parameters determining the whole road alignments are resolved simultaneously,and the indicators estimating the accuracy of digitized curve features are obtained as well.In the practical situation,the road curves are usually digitized from piece to piece,therefore,the practical adjustment models are further derived to process integrally the segmentally digitized road curves.Case studies are presented to illustrate the application of the models in the adjustment of a stretch of digitized road curve composed of the basic geometric features. By using developed models and methods for digitized road curve adjustment,the error influences on the road curve digitization are minimized and the accuracy of the digitized road curves is improved.The accuracy of the digitized road curve features is estimated with road curve adjustment method,as provides the quality indicators for the spatial databases of GIS-T.On the other hand,the system parameters of the digitized road curves obtained by using the developed models are very close to those of the designed road curves,as is the blocks for further development of the datum-based linear referencing system and geometric curve feature-oriented dynamic segmentation model for GIS-T.