The outcome in adolescence of children diagnosed as dyslexic during the early years of school was examined in children prospectively identified in childhood and continuously followed to young adulthood. This sample offers a unique opportunity to investigate a prospectively identified sample of adolescents for whom there is no question of the childhood diagnosis and in whom highly analytic measures of reading and language can be administered in adolescence. Children were recruited from the Connecticut Longitudinal Study, a cohort of 445 children representative of those children entering public kindergarten in Connecticut in 1983. Two groups were selected when the children were in grade 9: children who met criteria for persistent reading disability in grades 2 through 6 (persistently poor readers [PPR]; n = 21) and a comparison group of nondisabled children, subdivided into average readers (n = 35) and superior readers (n = 39). In grade 9, each child received a comprehensive assessment of academic, language, and other cognitive skills. Measures of phonological awareness (but not orthographic awareness) were most significant in differentiating the 3 reading groups, with smaller contributions from measures of word finding and digit-span. Academic measures that best separated good from poor readers were decoding and spelling, whereas measures of math and reading comprehension did not. Measures of phonological awareness, followed next by teacher rating of academic skills were the best predictors of decoding, reading rate, and reading accuracy. In contrast, the best predictor of reading comprehension was word finding, with digit span and socioeconomic status also contributing significantly. Using a growth curve model (quadratic model of growth to a plateau) all 3 groups demonstrated similar patterns of growth over time, with the superior group outperforming the average group, and the average group outperforming the PPR group. There was no evidence that the children in the PPR group catch up in their reading skills. Deficits in phonological coding continue to characterize dyslexic readers even in adolescence; performance on phonological processing measures contributes most to discriminating dyslexic and average readers, and average and superior readers as well. These data support and extend the findings of previous investigators indicating the continuing contribution of phonological processing to decoding words, reading rate, and accuracy and spelling. Children with dyslexia neither spontaneously remit nor do they demonstrate a lag mechanism for catching up in the development of reading skills. In adolescents, the rate of reading as well as facility with spelling may be most useful clinically in differentiating average from poor readers.