Selective cell fusion is a natural part of development. It is found in sexually reproducing organisms that require fertilization to propagate and in muscles, placenta, bones, lens of the eye and stem cells. Cell fusion is particularly important in the development of C. elegans: in addition to 300 sperm and oocytes that fuse during fertilization, 300 of the 1090 somatic cells born, fuse throughout development. Studies of cell fusion in C. elegans have shown that although different types of cells fuse, cell membrane merger is initiated through a common mechanism involving the action of one gene, eff-1. In worms with mutations that inactivate eff-1, almost none of the 300 somatic cells that normally fuse do so, but appear to differentiate, attach and behave in the same way as fusing cells. Such worms develop and survive but have numerous morphological, behavioral and fertility defects associated to cell fusion failure in the epidermis, pharynx, male tail, vulva and uterus. Cell fusion in embryonic dorsal epithelial cells has been analyzed in great detail by confocal microscopy using membrane fluorescent probes, apical junction markers and cytoplasmic aqueous fluorescent probes allowing the direct observation of membrane disappearance, pore expansion and cytoplasmic content mixing. The complete elimination of the membranes between two fusing cells takes about 30 min and involves vesiculation of the fusing membranes. Genetic and cell biological evidence indicates that eff-1 activity is both necessary and sufficient to fuse epithelial and myoepithelial cells in vivo. Based on electron microscopic analyses of intermediates of cell fusion in eff-1 mutants, it appears that eff-1 is required for both initiation and expansion of fusion pores, similar to the fusogen of Influenza virus. While only one gene encoding a novel candidate component of the cell membrane fusion machinery has been found, the nematode's cell fusion program is under the control of many cell-specific transcriptional regulators. A large number of these conserved regulators prevent cell fusion by repressing eff-1 activity. For example, if either ceh-16/engrailed or the GATA factor EGL-18/ELT-5 is inactivated, the lateral epidermal cells that normally do not fuse in the embryo will fuse causing embryonic lethality. And if either the Hox protein lin-39/Deformed or its cofactor ceh-20/Extradenticle is inactivated, the ventral epidermal vulval precursor cells that normally do not fuse in the larvae will fuse and the hermaphrodite will have no vulva. In addition, there is evidence for coordinated and complex regulation of lin-39 in the ventral epidermis by Ras, Wnt, Rb/E2F, NuRD and lin-15 pathways. It appears that in many cells that normally do not fuse, specific transcription complexes repress eff-1 expression preventing cell fusion. ref-2 (REgulator of Fusion-2) encodes a Zn-finger protein that is required to generate ventral Pn.p cells and to keep them unfused both in males and hermaphrodites. ref-2 is necessary, but not sufficient, to maintain Pn.p cells unfused. This review shows that far from cell fusion being an unusual phenomenon, there is the clear prospect that animal cells in all tissues are intrinsically programmed to fuse, and are only prevented from fusing by transcriptional and post-transcriptional control mechanisms. There are three major questions that remain open for future research: (1) How does eff-1 fuse cells? (2) How do Ras, Wnt, Rb, NuRD, E2F, heterochronic and other pathways control cell fusion? (3) What are the implications of cell fusion beyond worms?