In spatial data integration the most difficult problems arise due to multiple, inconsistent representations of one and the same real world object in different geospatial databases. One of the biggest challenges regarding the integration of multiple representations is the identification of corresponding objects within diverse source data sets. This process is generally referred to as spatial data matching. Some sophisticated approaches have been presented to cope with the task but still methods are needed to optimize the procedure. In this work, it is intended to achieve such an optimization by applying an iterative approach for the matching of street data form disparate sources, namely the Geographic Data Files (GDF) format and the German Authoritative Topographic Cartographic Information System database (ATKIS). After reducing the global geometric deviation of the linear source data sets by a rubber sheeting transformation, the street objects are topologically split and additional nodes are introduced, respectively, in order to enable the detection of a maximum number of 1:1 matches. Then, the matching process starts by identifying seed nodes in the source data sets which show a high likelihood of correspondence. With the seed nodes as starting points, a combined edge and node matching algorithm detects 1:1 correspondences. In case no 1:1 match could be found, an enhanced edge matching approach being able to recognize 1:2 matches is triggered. The whole process is performed in multiple iterations and it is repeated applying relaxed constraints. The results of the matching are stored in explicit relations expressing the degree of inconsistency of multiple representations.