Background The personality dimensions of harm avoidance (HA) and self-directedness (SD), as measured by the Temperament and Character Inventory (TCI), have been widely associated with depression and there is preliminary evidence that they may represent trait markers for depression. However, many studies in this area are limited by the use of heterogeneous samples of depressed patients and by the confounding effect of depressed mood during personality testing. The current study compares TCI personality dimension scores in a group of euthymic young adults with recurrent early-onset major depressive disorder (RE-MDD) to well-matched euthymic controls. Methods Fifty-two young adults with a past history of RE-MDD were recruited from consecutive referrals to a psychiatric clinic at a university health service. Eighty nine controls were also recruited. Euthymia was established in patients by a score of less than 9 on the Hamilton Rating Scale for Depression (HRSD) and in controls by a Becks Depression Inventory (BDI) score of less than 10. All participants completed the TCI-125. Results Patients and controls were well matched in terms of sociodemographic profile. Euthymic RE-MDD patients scored significantly higher than controls on the temperament dimension of harm avoidance (HA; mean score 14.5 versus 7.8, p < 0.0001) and significantly lower than controls on the character dimension of self-directedness (SD; mean score 14.1 versus 19.9, p < 0.0001). Covariance analysis suggested that both HA and SD contributed independently to the familial risk of depression. Limitations Subjects and controls all came from relatively affluent social backgrounds—these findings may not generalise to more socioeconomically diverse populations. The possibility of a ‘scarring effect’ of depressive episodes on self-reported personality dimension scores cannot be excluded. Conclusions High HA and low SD represent trait markers for liability to recurrent major depressive disorder in young adults. Further research is needed to replicate these findings and to assess the contribution that the experience of depressive episodes makes to self-reported personality dimension scores.