Depression has strong negative impacts on how individuals function, leading to the assumption that there is strong negative selection on this trait that should deplete genetic variation and decrease its prevalence in human populations. Yet, depressive symptoms remain common. While there has been a large body of work trying to resolve this paradox by mapping genetic variation of this complex trait, there have been few direct empirical tests of the core assumption that there is consistent negative selection on depression in human populations. Here, we use a unique long-term dataset from the National Health and Nutrition Examination Survey that spans four generational cohorts (Silent Generation: 1928–1945, Baby Boomers: 1946–1964, Generation X: 1965–1980, and Millenials: 1981–1996) to measure both depression scores and fitness components (lifetime sexual partners, pregnancies, and live births) of women from the United States born between 1938–1994. We not only assess fitness consequences of depression across multiple generations to determine whether the strength and direction of selection on depression has changed over time, but we also pair these fitness measurements with mixed models to assess how several important covariates, including age, body mass, education, race/ethnicity, and income might influence this relationship. We found that, overall, selection on depression was positive and the strength of selection changed over time–women reporting higher depression had relatively more sexual partners, pregnancies, and births except during the Silent Generation when selection coefficients neared zero. We also found that depression scores and fitness components differed among generations—Baby Boomers showed the highest severity of depression and the most sexual partners. These results were not changed by the inclusion of covariates in our models. A limitation of this study is that for the Millenials, reproduction has not completed and data for this generation is interrupted by right censoring. Most importantly, our results undermine the common belief that there is consistent negative selection on depression and demonstrate that the relationship between depression and fitness changes between generations, which may explain its maintenance in human populations.