To ascertain the direction of causality and differences by sex between major depressive disorder (MDD) and labor market outcomes in the US population, we used structural equation models separately for males and females to assess prospectively the interdependency of depression and labor market outcomes at Waves 1 (2001-2002) and 2 (2004-2005) of the National Epidemiologic Survey on Alcohol and Related Conditions. Diagnosis of MDD used DSM-IV criteria. We found that MDD at Wave 1 predicted being out of the labor force for males at Wave 2 (p = 0.006) and being in the labor force at Wave 1 was associated with lower probability of MDD at Wave 2 (p = 0.049). Among males in the labor force, MDD at Wave 1 was negatively associated with employment at Wave 2 (p = 0.047), and employment at Wave 1 was negatively associated with MDD at Wave 2 (p < 0.001). For women, there was no association between MDD and labor force participation. However, among women in the labor force, MDD at Wave 1 was negatively associated with employment at Wave 2 (p = 0.013) and being employed at Wave 1 was negatively associated with MDD at Wave 2 (p < 0.0001). These results indicate that MDD and negative labor market outcomes are associated with one another at both time points, but the effects differ by sex. To reduce the economic and social burden of MDD, these differences should be considered in clinical practice, vocational rehabilitation, and in the design of social policies.