Aim: A deficit in non-word repetition (NWR), a measure of short-term phonological memory proposed as a marker for language impairment, is found not only in language impairment but also in reading impairment. We evaluated the strength of association between language impairment and reading impairment in children with current, past, and no language impairment and assessed any differential impairment of NWR, compared with two other tests of verbal memory in children with language impairment with and without reading impairment. Method: Our sample comprised children aged 6-16y 11mo participating in a study of the genetics of language impairment: 78 children from 68 families (53 males, 25 females) with current language impairment (C-LI), compared with their 74 siblings: 25 children (18 males, seven females) with a past history of language impairment and 49 children (27 males, 22 females) who had never had a language impairment. The tests used were the Clinical Evaluation of Language Fundamentals (CELF III), the Children's Test of Non-word Repetition (CN-Rep), the Wide Range Assessment of Memory and Learning (WRAML) verbal memory index, the Wechsler Intelligence Scale for Children-III (WISC-III) digit span, and the Wechsler Objective Reading Dimensions (WORD[superscript UK]). Results: Reading impairment was present in two-thirds of the children with current language impairment. NWR deficits were significantly worse in children with language impairment who had reading impairment in reading decoding (p = 0.007 and 0.004--average group compared with borderline and definitely impaired groups respectively) or spelling (p = 0.002 and 0.005--average group compared with borderline and severely impaired groups respectively) (not correlated with severity of language impairment) but not comprehension impairment. In contrast, WISC digit span and WRAML verbal memory were impaired in all children with language impairment and did not differentiate those who also had reading impairment. Interpretation: We suggest that current NWR ability may be a marker of a process specifically underlying language impairment, co-occurring with reading impairment involving reading decoding and spelling, rather than a generic correlate of language impairment. Other verbal memory deficits appear to be pervasive in children with language impairment.