Controversy exists concerning the reliability of retrospective self-reports on childhood abuse since this method might be subject to under- or overreporting. Until now, no studies have been done in older adults, although reasons for under- or overreporting could be even more prominent in this age group. In this first study in older adults, test-retest reliability of retrospective measurements on childhood abuse and the influence of age, cognitive functioning and depression on this test-retest reliability was investigated. A longitudinal cohort study, the Netherlands Study of Depression in Older persons (NESDO), obtained information on childhood abuse at baseline and at a 6-year follow-up interview. Our sample consisted of 277 adults (mean age 68.5 years at baseline) of which 118 (42.6%) reported childhood abuse at baseline. The largest proportion of the answers was consistent (yes-yes or no-no) for every type of childhood abuse, varying from 85.2 to 93.5%. Looking more closely, ‘yes’ answers were more fluctuating than ‘no’ answers. Sexual abuse was most reliably reported in two separate interviews. There was no significant effect of age nor cognition on the test-retest reliability. Only test-retest reliability of emotional neglect was significantly associated with depression diagnosis and depression severity at baseline. In conclusion, test-retest reliability of retrospective self-reports on childhood abuse seems moderate to good, but slightly dependent on the type of abuse. The test-retest reliability was influenced neither by age nor cognition, only reporting of emotional neglect was associated with depression state.