At present, it has become a hot issue to provide a multi-representation mechanism and build multi-scale spatial databases in the field of GIS. Meanwhile, an inevitable problem is how to evaluate and preserve the consistency of multi-scale spatial data. In this paper, we introduce the concepts of homonymous points and homonymous entities based on innate characteristics of spatial data and put forward a seamless constraint model of multi-scale representation of geographic information that involves time, geometry, spatial relations and semantic characteristics. According to this model, a consistency assessment method has been implemented by computing similarities between vector data at three different scales. The main contributions of this paper are the proposal of a consistency constraint model and assessment system of multi-scale spatial data.