If one parent is not genotyped then bias can occur if affected offspring who are homozygous for the marker are included in the analysis, but for a multiallelic marker heterozygous offspring can be used.
If neither parent is typed then biases can again arise, even if it is possible to unambigously deduce the parental genotypes.
Probably more important than the fact that bias can occur when reconstructing missing parental genotypes is the fact that much useful information is discarded when such reconstruction is not possible. This is especially the case when the disease polymorphism itself is studied, since then one may be dealing with a biallelic system and often the parent bearing the susceptibility allele may be the one who is missing.