Objectives: The current study aimed to describe and predict perinatal depression trajectories in a sample of low-income women from the first trimester of pregnancy to six months postpartum. Methods: The study sample consisted of 899 women in Wisconsin who received home visiting services. Eligible participants were screened for depressive symptoms by home visitors using the Edinburgh Postnatal Depression Scale at least three times across four time periods from the first trimester of pregnancy to six months postpartum. Growth Mixture Modeling was applied to identify distinct trajectory classes, and multinomial logistic regressions were performed to analyze predictors of class membership. Results: Mean depressive symptom scores in this racially/ethnically diverse sample of low-income women decreased significantly over time from a high of 8.1 at time1 to a low of 6.8 at time4. Four classes were identified, including a low-stable group (78.2% of sample), a high-stable group (10.6%) along with decreasing (7.1%) and increasing (4.1%) trajectories. Women with a history of abuse and mental health difficulties were more likely to be classified in the high-stable and decreasing depression groups than the low-stable group. Low social support was linked to an increasing trajectory that resulted in high levels of postpartum depression. Conclusions and implications: Although most women exhibited stable and positive trajectories, more than one out of five presented with either persistently or intermittently high depression scores. Taken together, the findings underscore the importance of depression screening throughout the perinatal period and identifying factors that may be used to target resources to at-risk populations.