Depression is a serious mental health issue. Much research has been devoted to understanding and treating it. Despite this, we still have limited understanding of how certain factors such as socioeconomic status, social support, and childhood abuse affect the likelihood of recovery. Using survey data from the Midlife in the United States series, we develop logistic and tobit models to evaluate the impact of health, social, and perception-based factors on depression severity and the odds of recovering. We stratify our sample by both initial depression severity and partnership status. Our results suggest both depression severity and partnership status influence which factors are significant and insignificant in determining who recovers from depression.