To determine the natural history of developmental dyscalculia (DC) and factors impacting on its prognosis, we performed a prospective six-year longitudinal study. One hundred and forty children of normal intelligence diagnosed with DC in the fifth grade of elementary school were re-examined for dyscalculia three and six years later, in eighth (n=123) and eleventh (n=104; 41 males, 63 females) grades respectively. Mean age of the children in fifth grade was 11 years 1 month (SD 4 months), in eighth grade 14 years 2 months (SD 1 month), and in eleventh grade 17 years 2 months (SD 5 months). The assessment included standardized arithmetic, reading and writing tests, behavioural rating scales, information on socioeconomic status, educational interventions, and familial learning problems. Participants in eleventh grade were recategorized as having DC if their score on the arithmetic test was not more than the fifth centile for grade. At the six-year follow-up, 99/104 (95%) children diagnosed with dyscalculia in fifth grade were still performing poorly in arithmetic, scoring within the lowest quartile for their grade, and 42/104 (40%) were recategorized with DC. Chronicity of DC was associated with severity of the dyscalculia in fifth grade (p<0.05), lower IQ (p<0.01), inattention (p<0.01), and writing problems (p<0.01). Thus, DC is an enduring specific learning difficulty, persisting into late adolescence in almost half of affected individuals.