Rising research costs, broadening goals, intellectual property rights, and other concerns increase the need for robust management of crop improvement data. The data model of the International Crop Information System (ICIS) allows breeding processes to be recorded unambiguously in a relational database. This paper describes this model, which underlies the Genealogical Management System (GMS) of ICIS. The model recognizes three classes of methods by which genetic material is advanced. Generative methods such as crossing or mutagenesis increase variation. Derivative methods usually involve selection, and maintenance methods conserve the genetic makeup of germplasm, such as in seed multiplications. Unlike systems that only track pedigrees, the model describes steps of selection. Applications are illustrated for self-pollinating, outcrossing, and clonally propagated crops. The ICIS GMS is in use for species including rice (Oryza sativa L.), wheat (Triticum aestivum L.), maize (Zea mays L.), potato (Solanum tuberosum L.), common bean (Phaseolus vulgaris L.), lesquerella [Lesquerella fendleri (Gray) S. Wats.], and witloof chicory (Cichorium intybus L.). The International Rice Information System, based on ICIS, holds more than 2.6 million unique identifiers for germplasm accessions, crosses, populations, and lines, requiring about 900 megabytes of storage space, which can easily be managed on a personal computer. The GMS model appears suited for widespread use in managing data on crop improvement. Crop Science Society of America.