Ploidy manipulations and interspecific crosses have allowed considerable genetic progress in Musa breeding, but estimation of genetic parameters for parental selection remains a major challenge. This study aimed to determine the components of genetic variance and the relative contribution of genetic effects to phenotypic variation of yield and phenological traits in secondary triploid hybrids from tetraploiddiploid crosses. The hybrids were evaluated in two consecutive growth cycles on a tropical forest site. Non-genetic effects accounted for a large fraction of the variation observed for most traits, except bunch weight. Partitioning of genetic variance into additive and dominance components confirmed the predominant role of additive genetic effects on the expression of bunch weight, fruit filling time, fruit length, plant height, and number of leaves while primarily non-additive effects accounted for suckering behavior and fruit circumference. Maternal general combining ability (GCA) effects accounted for the additive genetic variation in plant height and number of leaves, suggesting that selection for these traits should be carried out in tetraploid clones. Conversely, paternal GCA effects were the primary causes of genetic variation for fruit filling time, bunch weight, and fruit length, suggesting that these characters should be selected for in diploid clones. Specific combining ability (SCA) effects were observed for all traits, except fruit filling time, suggesting that additional genetic gain could be achieved through recombinative heterosis for these traits. \n