This paper answers the question as to which parental phenotypes are useful in prediction of the linkage phase of the progeny under conditions of linkage equilibrium. I demonstrate that 1) the parents with the same disease status or marker genotype carry no information about phase of the diheterozygous offspring; 2) the linkage phase can be inferred unambiguously only under recessive or dominant control of disease. In any other situations, only the probability of linkage phase is inferred. It has been shown that the relative difference between the probabilities of coupling phase and repulsion phase, epsilon, which is regarded as a measure of accuracy of linkage phase prediction by parental phenotypes, is the same in all informative crosses and is dependent on the genetic model of the disease. Furthermore, epsilon is the same for the di- and polyallelic marker genes.