We were intrigued by the pedigrees in the paper by Ceroni et al1 in the October 2014 issue that explored the association between ZNF277 microdeletions and specific language disorder (SLI). The authors make the case for ZNF277 microdeletions contributing to the risk of SLI while acknowledging that there was little evidence of co-segregation of these microdeletions with SLI in their discovery pedigree plus eight other families. The authors only discuss this discrepancy in terms of reduced penetrance, that is, unaffected siblings that carry the microdeletion, but say little about the reverse discordance in two of their three multiplex families, that is, where one of the two affected siblings has inherited the maternal microdeletion and the other has not. This reverse discordance has been reported in multiplex families with autistic spectrum disorder (ASD) in which a putative risk copy number variation (CNV) is segregating.2 This paper on ASD, and our finding of a maternal grandmother age effect in ASD prompted us to propose the meiosis mismatch methylation (3M) hypothesis.3 This states that, in female early meiosis I, the pairing of a chromosome carrying a microdeletion with a wild-type homologue increases the chance of abnormal methylation due to chromosome looping through misaligned pairing, such as would occur with silencing of a transposon.4 This in turn results in the functional silencing of the wild-type gene when transmitted to offspring. 3 M predicts that all offspring of a woman carrying a risk CNV are at risk of the condition even though only half inherit the CNV. Independently, around the time of formulating the 3 M hypothesis, an ASD family was reported in which one affected sibling had inherited a maternal deletion containing the oxytocin receptor gene (OXTR) and the other affected sibling had no deletion, but had epigenetic misregulation of this gene through aberrant gene silencing by DNA methylation.5 We suggest that it would be useful to examine at least the DNA methylation status of the maternally inherited ZNF277 allele in the siblings that have SLI but not the microdeletion.