Abstract Inheritance of sex forms in watermelon ( Citrullus lanatus ) is not well described. In this study, we made five pairs of crosses of watermelon plants with different sex forms and grew progeny in two different seasons to investigate the inheritance of sex forms and the seasonal effect on sex expression. We showed that environmental factors have no effect on sex forms, but they affect sex expression on individual flowers as more pistillate flowers were observed in spring than in autumn. This suggests that short photoperiod and low temperatures promote formation of pistillate flowers in watermelon. In the F 2 population of the cross of andromonoecious (SL3H or AKKZW) × monoecious (XHB), the segregation ratio is 9 monoecious: 3 trimonoecious: 4 andromonoecious, and the segregation ratio in BC 1 P 1 (F 1 × andromonoecious parent) is 1 monoecious: 1 trimonoecious: 2 andromonoecious. The segregation ratio in the F 2 population of the gynoecious (XHBGM) × monoecious (XHB) is 3 monoecious: 1 gynoecious whereas the segregation ratio in the BC 1 P 1 (F 1 × gynoecious parent) is 1 monoecious: 1 gynoecious. The segregation ratio in the F 2 population of gynoecious × andromonecious cross is 27 monoecious: 12 andromonoecious: 9 gynoecious: 9 trimonoecious: 4 hermaphroditic: 3 gynomonoecious. The segregation ratio in the BC 1 P 1 population (F 1 × gynoecious) is 1 monoecious: 1 gynoecious whereas the segregation ratio in the BC 1 P 2 (F 1 × andromonoecious) is 1 monoecious: 1 trimonoecious: 2 andromonoecious. Taken together, the results suggested that three recessive alleles, andromonoecious ( a ), gynoecious ( gy ) and trimonoecious ( tm ) control the sex forms in watermelon, and a allele is epistatic to the tm allele. The following phenotype-genotype relationships are proposed for each of the sex forms in watermelon: monoecious, A_Gy_Tm_ ; trimonoecious , A_Gy_tmtm ; andromonoecious, aaGy_Tm_ or aaGy_tmtm; gynoecious, A_gygyTm_ ; gynomonoecious, A_gygytmtm ; and hermaphroditic, aagygyTm_ or aagygytmtm .