Abstract Previous studies have shown that high cognitive ability, measured in childhood and prior to the experience of traumatic events, is protective of PTSD development. Our aim was to test if the association between pre-trauma verbal ability ascertained at 5 years with DSM-IV lifetime post-traumatic stress disorder (PTSD) at 21 years was subject to effect modification by gender, trauma type or prior behaviour problems. Using a prospective birth cohort of young Australians, we found that both trauma type and behaviour problems did not change the association between cognitive ability and PTSD. During multivariate analysis, testing for the interactive effect of gender revealed that verbal ability was linearly and inversely associated with PTSD in females only, with those in the lowest verbal ability quintile having strongly increased odds of PTSD (OR=3.89: 95% CI; 1.50, 10.10) compared with those in the highest quintile. A graph of the interaction revealed lower verbal ability placed females, but not males, at an increased risk of PTSD. Our results indicate that lower verbal ability in early childhood is a vulnerability factor for PTSD in females but not in males, and may constitute a gender-specific risk factor responsible for part of the increased risk of PTSD found in females compared with males.