Most recent epidemiologic studies investigating the potential health effects of occupational magnetic field (MF) exposure have relied on MF measurement data linked to job titles. These measurements are summarized by occupational categories, which represent similar groups of job titles. However, job titles alone explain only a small proportion of exposure variability. A comprehensive MF occupational exposure database was used to (1) develop summary job-specific estimates of magnetic field exposure, (2) evaluate the impact of incorporating work environment data to improve electric and magnetic field exposure assessment, and (3) evaluate the use of random versus nonrandom sampling when estimating mean MF exposure levels by occupational categories. Uniform classification systems were developed for occupational and work environment data. A factorial design was used to summarize and calculate arithmetic means and 95% confidence intervals for occupational MF data, assuming that the total variation in MF exposure resulted from variation in occupation, work environment, utility, worker, and day. Occupation-specific means varied across different work environments, particularly for craft workers. Although within-worker and between-worker variability account for a large proportion (over 50%) of exposure variation, work environment (24%) accounted for more exposure variability than occupation (4.9%) or utility (15%). Some differences were observed when results were compared from surveys that used random and nonrandom sampling; however, these differences were not consistent or systematic. It was concluded that MF exposure assessment should consider work environment in addition to job title to reduce exposure misclassification.