A complete and correct database is fundamental to reliable applications of GIS. Due to the nature of geographic data and limitation of measurement, all geographic data are of limited accuracy. In order to assure the credibility of GIS analysis, the quality of data is an important issue within GIS research. Overlay is a common operation of GIS, which is very sensitive to the quality of data involved. As a result, besides of its widely use in site selection and change detection, overlay is often adopted for error detection of GIS data. This research uses overlay as a tool to investigate errors in two steps. First, a number of erroneous map layers are created using the S-Plus software and ARC/INFO. These maps contain errors of various types and magnitude. Through overlay, this research investigates the relation between data quality and the results of overlay. Second, this research uses spatial factors, such as acreage and shape index, to identify and eliminate sliver polygons. Results show that slivers are not thin and long as we used to think. This controversial finding is subject to further research.