Little is known about the long-term mental health of women following the birth of an infant. This study describes the 21 year trajectory of women's depression following the birth of an infant and identifies early predictors of post-birth maternal depression trajectories. The sample comprises 2,991 women from the Mater and University of Queensland Study of Pregnancy. Using the Delusions-Symptoms-States-Inventory, depression was measured at 6 months, 5, 14 and 21 years after the birth. These measures were clustered and in addition bivariate and multivariate analyses were used to test for significant association between the groups and a range of maternal socio-demographic, psychological and pregnancy-related factors. Two depression trajectories were produced, a no-low depression group (79.0 %) and a high-escalating depression group (21.0 %). The strongest predictors for a high-escalating depression group were conflict in the partner-relationship (p < 0.001), anxiety (p < 0.001) and stress (p < 0.001) in the antenatal period, having many pregnancy symptoms (p < 0.001), being younger (p < 0.001) and having poorer social networks (p < 0.001). To a lesser extent not completing high school (p < 0.05), being unsure about wanting the pregnancy (p < 0.05) and not wanting contact with the infant following the birth (p < 0.05) were also predictors for high-escalating depression trajectory. Our findings suggest a sub-sample of mothers experience persistent depressive symptoms over a 21 year period following the birth of their infant. Partner conflict, inadequate social supports and poor mental health during the pregnancy, rather than factors relating to the birth event, contribute to women's depressive symptoms in the long-term. Given the identification of early markers for persistent depression, there may be opportunities for intervention for at-risk pregnant women.