The relationships between older age and sleep efficiency have traditionally been assessed using cross‐sectional studies that ignore changes within individuals as they age. This research examines the determinants of sleep efficiency, the heterogeneity in an individual's sleep efficiency trajectory across a period of up to 27 years in later life and its associations with health. The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort (n = 6,375; age 42–94 years) was used in this study. Depression and health data were collected using self‐report validated instruments (Cornell Medical Index, Beck Depression Inventory and Geriatric Depression Scale). Longitudinal sleep and sociodemographic data were collected using a study‐specific self‐report questionnaire. A mixed‐effect model was performed for sleep efficiency with adjustments for time‐invariant and time‐variant predictors. Latent class analysis was used to demonstrate subgroups of sleep efficiency trajectories and associations between sleep efficiency clusters and health history of the participants were investigated. Older adults have decreased sleep efficiency over time, with 18.6% decline between 40 and 100 years of age. Three sleep efficiency trajectory clusters were identified: high (32%), medium (50%) and low sleep efficiency (18%). Belonging to the high sleep efficiency cluster was associated with having lower prevalence of hypertension, circulatory problems, general arthritis, breathing problems and recurrent episodes of depression compared to the low efficiency cluster. Overall, ageing decreases sleep efficiency. However, there are detectable subgroups of sleep efficiency that are related to prevalence of different diseases.