Autosomal recessive and dominant inheritance of proximal spinal muscular atrophy (SMA) are well documented. Several genetic studies found a significant deviation from the assumption of recessive inheritance in SMA, with affected children in one generation. The existence of new autosomal dominant mutations has been assumed as the most suitable explantation, which is supported by three observations of this study: (1) The segregation ratio calculated in 333 families showed a significant deviation from autosomal recessive inheritance in the milder forms of SMA (P = .09 +/- .06 for onset at 10-36 mo and .13 +/- .07 for onset at > 36 mo; and P = .09 +/- .07 for SMA IIIa and .12 +/- .07 for SMA IIIb). (2) Three families with affected subjects in two generations are reported, in whom the disease could have started as an autosomal dominant mutation. (3) Linkage studies with chromosome 5q markers showed that in 5 (5.4%) of 93 informative families the patient shared identical haplotypes with at least one healthy sib. Other mechanisms, such as the existence of phenocopies, pseudodominance, or a second autosomal recessive gene locus, cannot be excluded in single families. The postulation of spontaneous mutations, however, is a suitable explanation for all three observations. Estimated risk figures for genetic counseling are given.