Objectives This study examined pathways through which job-related discrimination contributes to differential job control and influences health disparities.

Methods Data from the two MIDUS surveys (1995–2005) were used to examine JCQ scores for job control for black and white subjects, using a random-effects ANOVA model, controlling for demographic variables and education. Path analysis modelled the effect of discrimination on job control in 1995, and association with subsequent self-reported health in 2005.

Results Mean adjusted control scores in whites were higher than in blacks (32.2 vs 30.4; diff = 1.8, p=0.003). An interaction term for race and occupation did not reduce the difference between the two groups. Incorporation of two job discrimination measures into the model reduced the differential markedly (score difference 0.6; p=0.50, NS). Discrimination measures improved model fit and decreased the variance coefficient, indicating a substantive effect on assessment of job control. Job discrimination was associated with self-rated health in path analysis, and partially mediated its effect via job control in blacks, but not whites. Job discrimination was more likely in more educated black subjects (OR 1.13), and in those in higher-control jobs (OR 1.23).

Conclusions Individual job-related racial discrimination is an important mediator in the assessment of job control. This effect was independent of structural segregation into jobs. These results suggest that educational attainment in blacks is not matched by equivalent gains in job autonomy and authority. Prolonged diminution of job control in blacks may play a role in ‘weathering’ and erosive effects of discrimination on health.