A deterministic analysis is conducted to examine marginal dominance for two linked viability loci influencing inbreeding depression and its graphical inferences. Four estimators of marginal dominance are derived, assuming a biallelic marker locus completely linked to one of the viability loci, and the biases in expected estimates due to the other deleterious locus are discussed. Three conditions under which apparent partial dominance or underdominance could occur are found, i.e. when two multiplicative, partially recessive loci are linked in coupling phase and when two synergistic, highly overdominant loci are linked in coupling or repulsion phases. Expected frequencies of the three marker genotypes in selfed progeny are derived, considering two linkage phases, two types of marker locus position with respect to the viability loci, and the multiplicative and synergistic fitness models. Segregation ratios are generated for the marker locus linked to either two overdominant or partially recessive loci and plotted in gene action graphs to examine the robustness of the graphical inferences of gene action due to the presence of an additional linked viability locus. Under a multiplicative fitness model, the presence of an additional partially recessive or overdominant locus in the vicinity of the marker locus does not greatly affect the graphical inferences of the relative role of partially recessive or overdominant genes in expression of inbreeding depression. A marker linked to two synergistic, highly overdominant loci can behave as though linked to a partially recessive, partially dominant or underdominant locus, even with relatively weak synergism.