It is becoming common for soil surveys to be made of the same area at different intensities and published at different scales. The principles of cartographic generalization are discussed that control the relationships between the map units and delineations on maps made from such surveys. A study of two sets of maps showed that almost no lines were coincident. Up to 20% of the small scale delinations could be ’inliers’ of different soils and about 15% of the large scale delineations would be outside their small scale equivalents. The same discrepancies are to be expected between large scale soil maps and the smaller scale maps of physiography or vegetation that are often used to stratify soils. Reasons for these discrepancies are discussed under the headings of simplification and classification. Recommendations arc made to guide the preparation of maps and legends for different intensities and scales of survey in the same area. These recommendations have practical implications for the planning of surveys and the designs of computer-based autocartography systems.