Root-knot nematodes, Meloidogyne incognita in Guatemala and Meloidogyne sp. in EI Salvador frequently cause very serious damage to Coffea arabica. Hypocotyledon grafting on C. canephora var. Robusta is practiced on a very wide scale to control these pests. However, rootstock seeds come from non selected trees which provide 30-40% resistance. In this article, we examine the possibility of improving resistance to M. incognita and Meloidogyne sp. Hybrids were created using two North Carolina II factorial mating designs and tested under controlled conditions for resistance to isolates of the two nematodes. In the trial with Meloidogyne sp. the number of nematodes per plant was counted, and in the trial with M. incognita a root damage index was established. Vegetative measurements (height, weight of aerial parts and roots) were taken in both trials. The parents were classed according to their cross value and genetic variance was estimated. In the factorial trial conducted with the Meloidogyne sp. isolate from EI Salvador, parents T3561 and T3751 transmitted high resistance levels of 56 and 54%, respectively, to their progenies, as opposed to 9% for the other parents. The cross between those two parents achieved 78% resistant plants. In the trial with the M. incognita isolate from Guatemala, similar results were found. The same two parents transmitted resistance to 64% of their progenies, as opposed to 36% for the other parents. Classification of the parents did not differ from one trial to the other. The existence of a complex nematode resistance locus in the C. canephora species seems highly likely. The results show that it is possible to select rootstock varieties that are more resistant to the main Meloidogyne nematodes in Guatemala and El Salvador. Given the average heritability values (0.28-0.30) and the possibility of applying strong selection intensity, the genetic progress expected in the next selection cycle should be substantial.