To evaluate whether different evolutive profiles of depressive symptoms exist from pregnancy to 2-years postpartum, and whether early postnatal depressive symptoms (PNDS) predicted child developmental outcome independently of later maternal depressive symptoms, we studied data issued from a french prospective, longitudinal study, MATQUID (Quid of Maternity). Up to 598 mothers and children were followed-up from pregnancy to 2 years, with repeated measures of PNDS (Edinburgh Postnatal Depression Scale-EPDS; Center for Epidemiologic Studies Depression Scale-CESD) and of the children development (Bayley Scales II). Semiparametric mixture models have identified 4 distinct trajectories of maternal depressive symptoms (never depressed; depressed only during the postnatal period; depressed throughout the follow-up period; highly depressed throughout the follow-up period). Associations between 6 week PNDS, subsequent depressive symptoms and child development were evaluated with multilevel modelling techniques. Children of mothers with 6 week PNDS were significantly more likely than children of non- symptomatic mothers to have poor cognitive outcome, association reduced to trend level when adjusted for later maternal depressive symptoms. The main limitation of the MATQUID cohort is the low level of risk for PNDS, so these results cannot be extrapolated to all types of populations. Nevertheless, different subtypes of evolutionary profiles of PNDS have been found and the effects of early PNDS on infant development seem to be at least partly explained by subsequent depressive symptoms. These results emphasize the absolute necessity of using integrative models, both in research and in clinical practice, incorporating the evolution of parental mental health and the children development.