BACKGROUND: Sleep deprivation in adolescence is increasing in prevalence and may be linked to subsequent depression. Findings regarding associations between sleep duration, sleep onset time, and the development of depressive symptoms over time in adolescents are mixed, and rely on subjective measures of sleep. METHODS: Sleep onset and duration were assessed using a combined heart rate monitor and accelerometer and self-report in 688 participants from the ROOTS study at age 15. Participants reported depressive symptoms at ages 14.5, 16, and 17.5, using the Mood and Feelings Questionnaire. Latent growth curve modelling was used to model development of depressive symptoms and test associations with baseline sleep onset and duration. RESULTS: Cross-sectionally, falling asleep later and shorter sleep duration were both associated with higher depressive symptoms in males and females, using both device-measured and self-reported sleep. There were no longitudinal associations between baseline sleep duration and change in depressive symptoms. A later sleep onset-time was associated with a decrease in depressive symptoms over time, in females only, using device-measured sleep only. LIMITATIONS: The current sample was more economically advantaged and ethnically white than the UK average, and with lower MFQ symptoms than the original cohort, which may reduce generalisability. CONCLUSIONS: Adolescents who fall asleep later or sleep less have higher levels of depressive symptoms cross-sectionally, but do not show increases in depressive symptoms over time. Interventions targeting sleep onset and duration in adolescence may improve mental health in the short-term but the possibility of reverse causality should be explored further.