Objective: This study determined whether persons with major depressive disorder who switch or augment antidepressant therapy have higher health care costs and productivity losses than those who do not. Methods: Data from July 1, 2002, through June 30, 2006, were taken from a national employment-based medical and pharmacy claims database. Participants were required to have filled an antidepressant prescription, be treatment naïve six months before the index prescription, be continuously enrolled in the benefits plan at least six months before and 12 months after the index prescription, and have at least one outpatient-based medical claim for major depressive disorder. Participants were categorized according to whether they switched, augmented, or maintained (that is, neither switched nor augmented) their antidepressant therapy in the 12 months after the index prescription. Productivity losses were defined as days absent from work for medical visits multiplied by average daily wage. Multivariate analyses (generalized linear models) were used to compare costs per person in the year after the index prescription, and univariate analyses (Wilcoxon tests) were used to compare productivity losses per person. Results: Of the 7,273 individuals who met study criteria, 40.3% switched, 1.5% augmented, and 58.2% maintained the index antidepressant therapy. After the analyses controlled for baseline characteristics, mean total and depression-related health care costs, respectively, in the year after the index prescription were significantly greater for switchers ($9,288 and $1,388 per person) and for augmenters ($9,350 and $1,027) than for maintainers ($6,151 and $723). Mean total and depression-related productivity losses, respectively, were significantly greater for switchers ($2,081 and $680) and augmenters ($2,010 and $587) than for maintainers ($1,424 and $437). Conclusions: Persons with major depressive disorder who switched or augmented antidepressant therapy within 12 months of treatment initiation had higher resource costs and productivity losses than those who did not.