A North Carolina design II experiment, with three improved cassava ( Manihot esculenta
Crantz) accessions as the female parents, 15 cassava landraces and three improved cassava accessions as the male parents (3 x 18), was evaluated in three environments in Nigeria to determine the mode of gene action and the combining ability, and also to estimate heterosis for resistance to cassava mosaic disease (CMD), in various sources of resistance. General combining ability (GCA) effect due to females and males was significant in each environment; while the GCA effect due to males and the specific combining ability (SCA) effect were significant across environments. The relative magnitude of the total GCA components to the total GCA plus SCA component, however, suggested that GCA was more important that SCA in controlling CMD resistance among the crosses. The test for heterosis was significant in the individual environments; one across, involving the best general combiner, exhibited significant heterosis for resistance in all three environments. The implication of the findings in breeding for resistance to CMD is discussed.