For many geographic information system applications, such as pavement management and traffic inventories, sections of highway routes may be broken into predefined segments of fixed or variable lengths. The use of fixed-length segments or variable-length segments can present problems in data redundancy or require cumbersome spatial edits when attributes change. Dynamic segmentation was developed to address these problems. Used in conjunction with a milepoint linear-referencing scheme, dynamic segmentation can provide an efficient means for storing varying highway data. However, when dealing with very small segments and constantly changing attributes, the number of segments can make the use of dynamic segmentation slow and cumbersome. Introduced is a multicriteria dynamic segmentation approach that allows on-the-fly segment-length specification before a linear referenced table is attached to a route. Once the segment length is specified, the attribute data are binned by using database manipulation and aggregation operations based on a user-defined set of criteria. By using binned data, dynamic segmentation speed is enhanced considerably. Small-scale thematic maps of binned data are much more readable, and data spikes can be removed in the binning process. The underlying reference database containing the original small segment attribute data is never compromised in the binning process. The approach was developed specifically to accommodate managing pavement retroreflectivity data in a geographic information system. The system architecture is described, and other types of spatial data that may benefit from this approach are identified.