The genetic analysis undertaken here shows that the direct (i.e. proband) method for calculating risk figures is not readily applicable to von Recklinghausen neurofibromatosis (NF-1); the selection of available sibling groups for analysis becomes biased in various ways, primarily because of the wide phenotypic variation of the disease. However, indirect methods of analysis confirm that NF-1 shows autosomal dominant inheritance with full penetrance. The existence of an unusually high mutation frequency is also confirmed. In this study it is estimated to be between 4.3 x 10(-5) and 6.5 x 10(-5). However, in contrast to the findings of others, among sporadic cases, both their distribution within sibships and parental ages at delivery did not differ from random distributions. An assessment of the degree of severity of NF-1 and comparisons of the sporadic cases with the familial cases produced no evidence of any clinical somatic differences between the two groups, likewise for psychiatric evaluations of the two groups. Apart from 2 cases with non-NF-1 segmental forms of NF, it was not possible to distinguish alternative forms of NF among the sporadic cases. A pair of monozygotic twins with NF-1 is discussed with reference to the nature and localization of their respective tumours, which are not identical, indicating the influence of factors beyond the mutant NF-1 gene itself on the manifestations of the disease. In a genealogical study involving about 3,000 ancestors of patients from Gothenburg with known NF-1, families with common ancestors were not found, nor was it possible to demonstrate a tendency to clustering in one geographical area or isolated locality.