Objectives: To examine the prevalence of cognitive dysfunction among young men who had suffered a head injury during childhood or adolescence, in particular focusing upon the effects of age and the severity of the injury. Methods: By cross linkage of Danish national registers for hospital admissions and the draft board, 3091 young men were identified who had been injured before age 18 and tested at age 18 or shortly thereafter: 970 had suffered a single concussion and were in hospital for one day only; 521 had two concussions at separate times and were in hospital for one day only on each occasion; 961 had a cranial fracture; and 639 had a cerebral lesion. For all of these, scores on the draft board's cognitive screening test were obtained, dichotomised as dysfunctional or non-dysfunctional. Prevalences of dysfunctional scores were compared with population base rates (about 20% of all Danish men appearing before the draft board had a score classified as dysfunctional). Results: For young men who had suffered a single concussion, cranial fracture, or cerebral lesion before 12 years of age, resulting in less than 12 days of hospital admission (n = 376), rates of cognitive dysfunction did not exceed those in the general population (odds ratios < 1.4, NS), but the odds ratios became significant and > 1.4 where the injury occurred after age 11. For those with a cerebral lesion resulting in over 11 days of hospital admission (n = 263), all odds ratios were significant and > 2.0, irrespective of age at injury. For cases of two concussions, all odds ratios were > 1.4 but were not significant for all age groupings. Conclusions: For milder forms of single head injury before age 12 there is no evidence of enduring cognitive dysfunction. The apparent effect at later ages may reflect predisposing psychosocial factors. For more serious injury, and for cases with two concussions, the effect is marked and not clearly related to age. For the former, this suggests enduring neurological effects of the injury; for the latter, where the effect is unrelated to time between injuries, predisposing factors may again play a role.