Extensions to models originally described by Hartl for predicting the recurrence risk of new dominant mutations are developed in this paper. Additions to the models are (1) possible somatic mosaicism in a parent in some families, (2) the possibility that the parental origin of the mutation may or may not be known, and (3) mutation rates which change as a function of sex and/or time. The models indicate that recurrence risks are most critically affected by (a) the amount of somatic mosaicism which can be tolerated in the parent without manifesting disease and (b) knowledge of the parental origin of the mutation. In addition, there is a moderate effect on recurrence risks if mutation rates increase in the father as a function of time. Recurrence risks are at least as large as the risk of trisomy 21 in a child of a mother of age 35 years or older, probably much higher (5%-10%) when the mutation is known to be of maternal origin or if substantial somatic mosaicism in the parent is compatible with a normal phenotype. The recurrence risk of a new mutation is high because of a very high ascertainment bias of families with substantial germ-line mosaicism.