Epidemiological studies have reported process-specific elevated prevalence of beryllium sensitization (BeS) and chronic beryllium disease (CBD) among workers. However, exposure-response relationships have been inconsistent, possibly due to incomplete characterization of many biologically relevant aspects of exposure, including particle size. In 1999, two surveys were conducted 3-5 months apart at a beryllium metal, oxide, and alloy production facility during which personal impactor samples (n = 198) and personal 37-mm closed-face cassette (CFC) 'total' samples (n = 4026) were collected. Among process areas, median particle mass median aerodynamic diameter ranged from 5 to 14 μm. A large fraction of the beryllium aerosol was in the nonrespirable size range. Respirable beryllium concentrations were among the highest for oxide production [geometric mean (GM) = 2.02 μg m⁻³, geometric standard deviation (GSD) = 1.3] and pebbles plant (GM = 1.05 μg m⁻³, GSD = 2.9), areas historically associated with high risk of BeS and CBD. The relationship between GM 'CFC total' and GM respirable beryllium for jobs varied by process areas; the rank order of the jobs showed high overall consistency (Spearman r = 0.84), but the overall correlation was moderate (Pearson r = 0.43). Total beryllium concentrations varied greatly within and between workers among process areas; within-worker variance was larger than between-worker variance for most processes. A review of exposure characteristics among process areas revealed variation in chemical forms and solubility. Process areas with high risk of BeS and CBD had exposure to both soluble and insoluble forms of beryllium. Consideration of biologically relevant aspects of exposure such as beryllium particle size distribution, chemical form, and solubility will likely improve exposure assessment.