Background: Bishop and Snowling (2004) proposed that children with language impairments (LI) and children with reading difficulties (RD) can be considered to be on a (phonological) continuum of risk for reading impairments. Aims: The first aim of the present study was to address two specific hypotheses about the relationship between RD and LI. The severity hypothesis proposes that LI is a more severe form of RD. In contrast, the dyslexia-plus hypothesis proposes that children with LI share the same phonological deficit as children with RD, but also have impairments in non-phonological language skills. The second aim of the study was to further our understanding of the spelling skills of children with LI. Methods & Procedures: Twenty-three children with primary language impairments (LI) and 22 children with primary reading impairments (RD) matched on chronological and reading age were compared with typically developing children. The children's performance on tests of phoneme deletion, phonological memory, and spelling accuracy and phonetic spelling was compared. Outcomes & Results: Children with LI showed poorer phonological memory performance than children with RD and age controls, whereas both clinical groups performed significantly worse than the age controls on the phoneme deletion task and comparably to reading-age controls. Children with LI and RD were impaired for their age at phonetic spelling, the children with LI showing the more severe impairment. Further exploration of the data identified two different profiles among children with LI; some of these children had poorer phonological skills than predicted from receptive vocabulary, supporting the severity hypothesis. The other group showed a profile with concurrent impairments in vocabulary and phonological skills, in line with the dyslexia-plus hypothesis. Conclusions: The findings suggest that a subgroup of children with LI may be experiencing a severe form of dyslexia. The results further suggest that children with LI have difficulty producing phonologically plausible spellings. (Contains 1 figure and 2 tables.)