A statistical analysis of parental age and the incidence of new mutation has been performed. Some new data on Apert, Crouzon, and Pfeiffer syndromes is presented and combined with all available data from the literature on parental age and new mutation. Significant heterogeneity among syndromes for the rate of increase in incidence with parental age was found. A parsimonious conclusion is that mutations fall into two groups, one with a high rate of increase with age and the other with a low rate of increase with age. For the high-rate-of-increase group, a linear model relating incidence to age is rejected, while an exponential model is not. In addition, for this group, increased paternal age cannot account for the observed increase in maternal age--that is, increased maternal age also contributes to the incidence of new mutations. For the low-rate-of-increase group, increased paternal age alone can account for the observed increase in maternal ages; also, either a linear or exponential model is acceptable. In addition, there is no evidence for a mixture of parental age-independent cases with parental age-dependent cases for any of the syndromes examined. The curves reflecting incidence of new mutation and paternal age for two syndromes, Apert and neurofibromatosis, have an anomalous shape. In both cases the curve increases up to age 37 and drops at age 42 before increasing again at age 47. The usual explanation for the effect of parental age on new mutations is the mechanism of "copy-error" at mitotic division in male sperone that specifies an increased probability of mutation with time spent by a spermatozoon or ovum in a haploid state, a period of time that may also increase with age of the parent. A firm answer to the question of parental age and new mutation awaits identification of the molecular defect underlying some of these syndromes; we will then be in a position to determine in which parent the mutation occurred and at what age it did so.