The imaginal discs of Drosophila provide a favorable system for the analysis of the mechanisms controlling developmental cell proliferation, because of the separation in time between cell proliferation and differentiation, and the facility with which controlling genes can be identified and characterized. Imaginal discs are established in the embryo, and grow by cell proliferation throughout the larval period. Proliferation terminates in a regular spatial pattern during the final stages of larval development and the first day of pupal development. Cell proliferation can be locally reactivated in growth-terminated imaginal discs by removing part of the disc and culturing the remaining fragment in an adult host. The pattern of proliferation in these fragments suggests that cell proliferation in imaginal discs is controlled by direct interactions between cells and their neighbors. Proliferation appears to be stimulated by positional information differences, and these differences are reduced by the addition of new cells during tissue growth. Genes involved in cell proliferation control have been identified by collecting and analyzing recessive lethal mutations which cause overgrowth of imaginal discs. In some of these mutants (fat, lgd, c43, dco) the overgrowing tissue is hyperplastic; it retains its single-layered epithelial structure and is capable of differentiating. In two of the hyperplastic mutants (dco and c43), the imaginal discs show a failure of gap-junctional cell communication, suggesting that this form of cell communication may be involved in termination of proliferation. In other mutants the overgrowing disc tissue is neoplastic: it loses its structure and ability to differentiate, becoming a tumorous growth. The two genes that give a neoplastic phenotype (dlg and lgl) have been cloned and cDNAs of one of them (lgl) sequenced. The lgl gene encodes a cell surface molecule with significant homology to calcium-dependent cell adhesion molecules (cadherins). The expression of lgl at the time of termination of cell proliferation suggests that there are changes in the way that cells interact with one another at these times, and that these changes may be implemented by cell adhesion molecules. Direct cell contact within the epithelium, as well as signalling through gap junctions, appears to be involved in the cell interactions needed for the termination of cell proliferation. Mutations in genes encoding the Drosophila homologs of growth factors, growth factor receptors and oncogenes usually show an effect on cell-fate decisions rather than cell proliferation control, but this may be because oncogenic mutations in these genes would be dominant lethals and would therefore not be identified by conventional genetic analysis.