Study objectives: To explore sleep trajectories and identify the risk factors and mediators of poor sleep in middle-aged adults. Methods: Group-based multi-trajectory modelling was applied to the three waves of sleep data the from UK Biobank cohort to identify latent trajectories of sleep and group characteristics. Self-reported sleep duration, sleep problems (based on insomnia symptoms, snoring and trouble waking up) and daytime sleepiness (based on daytime tiredness and sleepiness) were included in the trajectory analyses. Multinomial logistic regression and mediation analysis were used to identify the main factors associated with poor sleep. Results: Analysis of sleep data from 41,094 participants (51.9% females) with a median age of 57 years (interquartile range 50–62 years) identified three distinct trajectories of sleep: healthy sleepers (40.8%); borderline poor sleepers (31.6%); and poor sleepers (27.6%). Socio-economic disadvantage, ethnic minority background, shift work, unhealthy lifestyle, poor health, depressive symptoms and obesity were the main risk factors associated with poor sleep. Around a third of the total effect of socio-economic deprivation on poor sleep was mediated through depressive symptoms. Conclusions: The distinct groups with differential risk for developing sleep issues and stable sleep trajectories highlight the non-transient nature of sleep issues. Early management of depressive symptoms can help in reducing the future burden of poor sleep. Due to the increased risk of poor sleep, people from socio-economically deprived groups, particularly females from ethnic minorities, should be the highest priority for interventions aiming to improve population sleep health.