Objective. We investigated maternal and paternal sleep evolution from 3 to 36 months postpartum, their interrelations and predictors in the SEPAGES cohort. Methods. Sleep information (night sleep duration [NSD], weekend daytime sleep duration [DSD] and subjective sleep loss [SSL]) was collected by self-administered questionnaires at 3, 18, 24 and 36 months postpartum in the SEPAGES French cohort that included 484 mothers and 410 fathers. Group-based multi-trajectory modelling was used to identify maternal, paternal and couple sleep multi-trajectory groups among 188 couples reporting sleep data for at least 2 time points. Multinomial logistic regression was used to assess associations between parental sleep multi-trajectories and early characteristics such as sociodemographic, chronotypes, child sex, birth seasonality or breastfeeding duration. Results. We identified three maternal (M1-M3), paternal (F1-F3) and couple (C1-C3) sleep multi-trajectory groups with similar characteristics: a group with short NSD and high SSL prevalence (M1, F2, C2), a group with long NSD but medium SSL prevalence (M2, F3, C3) and a group with long NSD and low SSL prevalence (M3, F1, C1). Mothers with the shortest NSD (M1) were less likely to have a partner with long NSD (F2). As compared with long NSD and low SSL prevalence (C1), couples with short NSD and high SSL prevalence (C2) were less likely to have had a first child born in the autumn and fathers in C2 had a later chronotype. Conclusion. We identified distinct sleep multi-trajectory groups for mothers, fathers and couples from 3 to 36-month postpartum. Sleep patterns within couples were homogeneous.