This paper reports a follow-up study of a cohort of 16- and 17-year-olds with a preschool history of speech-language impairment and whom Bishop and Edmundson (1987) originally studied. Information collected by questionnaire showed that the GCSE grades of those whose language impairments had resolved by 5;06 were below those of age-matched controls. However, the number of GCSE examinations entered and passed was significantly more than those of the 'persistent S-LI' and 'general delay' groups. Overall, IQ was the strongest predictor of educational attainment. However, even when IQ was controlled, literacy skills accounted for independent variance in achievement, especially among those with a history of language difficulty. The survey also noted that the majority of students across all groups remained in full-time education; however, the adolescents with a background of S-LI were more likely to follow vocational and employment training courses rather than A-levels.