Occupational agents were evaluated for the risk of brain-nervous system cancer in a cohort of 413,877 Finnish women with blue-collar occupations in 1970. Observed and expected numbers of incident cases and the intensities of exposure to 25 agents were generated for 183 job titles from 1971 to 1995. Poisson regression models linked incidence and exposure data. Increased risks were found for medium/high intensities of iron (standardized incidence ratio [SIR], 2.15; 95% confidence interval [CI], 0.96 to 4.80), oil mist (1.95; 0.97 to 3.90), any chromium compounds (1.51; 0.85 to 2.67), electromagnetic fields (1.37; 0.98 to 2.10), aliphatic and alicyclic hydrocarbon compounds (1.34; 0.80 to 2.27), lead (1.27; 0.81 to 2.01), cadmium (1.26; 0.72 to 2.22), and aromatic hydrocarbon compounds (1.20; 0.71 to 2.03). Strengths of the study include fair number of cases, virtually complete case coverage, and a high-quality job exposure matrix. Ecological design and cross-sectional job assessment introduced exposure misclassification and tended to drive risk estimates toward unity.