This paper considers the problem of integrating data sets which hold information about the same features in space. The approach proposed is based on allowing inconsistent data sets to coexist in the database and explicitly representing the nature of inconsistencies among those sets. A qualitative level of representation is created where the data sets can be integrated and manipulated using qualitative reasoning techniques. A systematic approach is used to identify classes and levels of consistency which can be checked in isolation. This provides the flexibility for two data sets to be integrated without necessarily being totally consistent in every aspect. The method is simple and can be used in many applications which incorporates the modelling and manipulation of spatial entities such as GIS, CAD and image databases.