A genetic analysis of resistance of cacao to Phytophthora palmivora was carried out in a 5 x 5 diallel and in a 4 x 2 NC II factorial design, involving cross-progenies and parental clones planted in the nursery and field, respectively. Resistance was scored in the laboratory by inoculation of leaf discs with P. palmivora spores with four replicates and, for the factorial design, results were compared with the percentage of rotten pods in the field assessed over a seven-year period. Significant differences between genotypes were observed for both laboratory and field evaluation methods. For the laboratory test, no reciprocal effect was observed and plant effects within seedling progenies were more significant than plant effects within clones. General combining ability was the main source of variation but specific combining ability was also significant for the leaf disc test applied to the diallel. Heritabilities increased with the number of replicates, reaching values of 0.34 and 0.67 for narrow sense heritability, and 0.60 and 0.67 for broad sense heritability, for the diallel and the factorial design, respectively. These values were higher than observed for the percentage of rotten pods in the field in the factorial design (0.42 and 0.47, respectively). For the leaf test, the expected genetic gains were around 30% with a selection intensity of 5%. The existence of significant genetic (0.71) and phenotypic (0.39) coefficients of correlation between resistance on leaves and percentage of rotten pods in the field in the factorial design confirms the feasibility of using the leaf disc test for early selection of resistance to Phytophthora pod rot of cacao.