<p>Good sleep quality is known to be associated with improved mental health, and vice-versa. However, the longitudinal association between mental health early in life and sleep later in life has not yet been rigorously investigated. The existence of such a relationship could imply that poor childhood mental health has lasting effects on sleep and, by extension, on physical health and mortality in the longer term. We used data from the 1970 British Cohort Study (<i>N</i> = 16 585). To operationalise childhood mental health, we separately analysed seven measures collected between the ages of 5 and 16. The outcome measures were self-reported, diary-derived and accelerometry-derived median nightly sleep duration at age 46. We conducted modified Poisson regression analyses with multiple imputation by chained equations. All seven measures of childhood mental health were positively associated with abnormal self-reported sleep duration. Five were associated with abnormal diary-derived estimates and four were associated with abnormal estimates from an accelerometry-based algorithm. Controlling for adult mental health symptoms post hoc attenuated the associations with self-reported sleep duration but not with the more objective estimates. This study provides clear evidence that poor mental health in childhood is associated with abnormal nightly sleep duration at age 46. Post-hoc analyses suggest that this effect might not be entirely mediated by mental health in adulthood. Presence of adult mental health symptoms may instead drive measurement error in self-reported sleep duration, and therefore result in biased estimates of the association with mental health earlier in life.</p>