The origin of the gland cells in asexually reproducing adult hydra is unclear. There is evidence suggesting that the gland cells are a self-renewing population as well as contrary evidence suggesting that they must arise from another cell type. We have reexamined the question and found the latter to be the case. Analysis of ectoderm/endoderm chimeras in which the ectoderm was labeled with [3H]thymidine indicates a precursor for gland cells in the ectoderm which migrates into the endoderm. Analysis of grafts between labeled lower halves and unlabeled upper halves of animals indicates the migratory precursor is either a large or a small interstitial cell. Measurement of the cell cycle times of the gland cells and the epithelial cells provided further support. The cell cycle time of the gland cells appears to be longer than that of the epithelial cells of the endoderm throughout the animal. This means that in the steady-state growth condition of hydra tissue, the gland cells cannot maintain their population size simply by cell division. These results and other data suggest the following dynamics for the gland cell population. Gland cells arise by differentiation from large interstitial cells, undergo a limited number of cell divisions, and then become postmitotic.