Somatic embryogenesis in cotton (Gossypium hirsutum L.) had been achieved as early as 1983, but only in certain cotton genotypes, especially 'Coker 312' and its sister lines. Somatic embryogenesis is important for the regeneration of intact plants from transgenic cells. Somatic embryogenesis has been suspected for some time to be genotype-dependant, but the mode of inheritance and the number of genes affecting the trait are unknown. In this research, crosses were made between four cotton genotypes; Coker 312 and TM-1, 'Stoneville 506' or 'Acala SJ-2'. Parents and their progenies (F1, F2, BC) were tested in tissue culture for somatic embryogenic potential. A complete randomized block design was used for tissue culture research, and calli were initiated from hypocotyl tissue of three to five day old seedlings. Globular embryos were present in calli after six weeks in culture, but many embryos could not be differentiated at that time. The embryos were transferred to culture flasks with liquid media until four weeks later, at which time they were scored. There was no evidence of maternal effects on somatic embryogenesis. Chi-square tests were used to test ratios corresponding to possible genetic systems for the embryogenic trait. Embryogenesis in Coker 312 was found to be controlled by one dominant gene. The expression of that dominant gene for embryogenesis is blocked by two or three dominant genes at separate loci in TM-1, Stoneville 506, and Acala SJ-2. Breeding for embryogenic potential in cotton is feasible, and would be best accomplished by a cycle of repeated backcrossing, selfing, and selection with Coker 312 as the donor parent. Large populations would be needed in the backcross generations of such a selection program because of the low frequency of embryogenic plants resulting from backcrossing.