In a previous report the authors described the organization of an aerial image analysis system. There are three levels of representation and control in that system: A High Level Expert (HLE) that utilizes a symbolic hierarchical model for the possible spatial organization of objects in the image to build partial, local interpretations of the image and to determine where to further analyze the image and what analyses to perform; a Model Selection Expert that determines, on the basis of contextual information provided by the HLE, the most promising appearance descriptions to use in searching for objects and structures in the image; and a Low Level Vision Expert that finds pictorial entities that satisfy these appearance descriptions by selecting image processing methods to find the appropriate entities. Emphasis has been on the High Level Expert, which is based on a general method of 'evidence accumulation' to perform flexible spatial reasoning. This paper contains a detailed description of the evidence accumulation process and its associated consistency checking process.