A job exposure matrix (JEM) was constructed for an epidemiologic study at an Ohio beryllium processing facility and was evaluated for temporal changes in airborne beryllium exposures. Quantitative exposure measurements of airborne beryllium were made between 1953 and 2006 and were used by plant personnel to estimate daily weighted average (DWA) exposure concentrations. We used these DWA measurements to create a JEM with 21 exposure metrics, which we linked to the plant cohort consisting of 18,568 unique job, department, and year combinations. The exposure metrics included quantitative metrics (annual average DWA exposures, maximum DWA and peak exposures), qualitative metrics (chemical beryllium species, and physical form), and qualitative assignment of exposure to confounding exposures. Twelve collapsed job titles with long-term consistent industrial hygiene samples were evaluated using analysis of variance (ANOVA) for time-trends in DWA estimates. The arithmetic DWA mean was 2.15 microg/m3and the geometric mean DWA was 1.60 microg/m3 for these jobs. After the DWA calculations were log-transformed, eleven of the twelve had a statistically significant (p < 0.001) decrease in reported exposure over time. ANOVA of all twelve jobs (analyzed as a group) indicated significant differences among years. Annual arithmetic mean DWA estimates for the JEM by decade ranged from a high of 1.39 microg/m3 in the 1950s to a low of 0.33 microg/m3 in the 2000s. The constructed JEM successfully differentiated beryllium exposures across jobs and over time. This JEM was combined with two others for use in a cohort study of workers at three beryllium processing facilities.