In order to examine the basis of familial associations with poor language achievement, 62 twinship pairs and 3 sets of triplets were studied in which at least one member presented poor oral language status in the absence of other developmental or sensory impairments. Rates of concordance for poor language were compared between the monozygotic (MZ) and dizygotic (DZ) pairs. Concordance for the MZ pairs was .96, whereas it was .69 for the DZ pairs. The DeFries-Fulker method for computing heritability of extreme scores was employed yielding a h2g of .45. Similar results were obtained for those pairs containing only children with performance IQs above 85, but with at least one member having language below -1 SD. A comparison of DZ twins with singleton sibling pairs with at least one affected member and of similar age showed that DZ twins were more similar with respect to language achievement than singleton pairs. Thus, twinning appeared to have influenced twin resemblance among those where at least one member presented poor language achievement. These results support the hypothesis that genetic factors contribute to the liability of poor language achievement in children who do not present other developmental or sensory disorders.