Research that has attempted to examine the relationship between work exposures to magnetic fields and health effects has suffered from limited exposure information. Power-frequency electric and magnetic (EM) field exposures vary substantially between industries, occupations, and individuals. In this study magnetic field data were collected for a variety of occupational categories within an electric utility. The sampling procedures emphasized craft occupations that were presumed to have higher exposures to magnetic fields. The objectives were to provide better exposure information for an ongoing cancer mortality study, examine the relationship between different summary measures of magnetic field exposure, and make available descriptive information useful for exposure reduction and worker education. Principal components analysis (PCA) and Varimax rotation were used to explore the relationships between the different summary measures among all utility workers and among the subset of electrical craft occupations. Discriminant analysis was used to assess summary measures of exposure that differentiated occupational groups. Measurements for 770 days generated a total of 42378 hours of magnetic field data. Eleven summary indices of exposure were calculated for specific occupational groups. These include arithmetic mean, geometric mean, median, 95th and 99th percentiles, and fraction of measurements exceeding .5, 1, 5, 10, and 100 microT. Electrical craft occupations had higher work exposures than non-craft occupations. Electricians and substation operators had the highest exposures among craft occupations. We identified subsets of summary measures that were intercorrelated. The first PCA axis included the geometric mean, median, and the fractions exceeding 0.5 and 1.0 microT. This set of measures also were best at discriminating occupational groups. The relevance of these findings may become more important if consistent associations are found between particular occupations and disease and particular occupations and magnetic field summary measures. Further research is needed to evaluate these exposure assessment findings.