Occupational exposure to extremely low-frequency magnetic fields (MF) was studied in 56 male subjects with breast cancer (adenocarcinoma) diagnosed in 1985-91, and 144 subjects with testicular cancer (seminoma and non-seminoma), diagnosed in 1985-87. The cases were compared with 1,121 control subjects from a previous case-control study on MF and cancer. Exposure assessment was based on the job held longest during the decade before diagnosis linked to a job exposure matrix based on MF measurements. The results refer to an estimated average mean of > 0.28 microT (Q4) and > 0.40 microT (P90, part of Q4) with < or = 0.15 microT (Q1) as reference. For breast cancer, the odds ratios (OR) and the 95 percent confidence intervals (CI) were 0.7 (CI = 0.3-1.9) and 0.7 (CI = 0.2-2.3), respectively. For men 60 years or younger, the corresponding estimates were OR = 0.9 (CI = 0.2-4.5) and 1.5 (CI = 0.3-8.3). For testicular cancer, the ORs were 1.3 (CI = 0.7-2.5) and 2.1 (CI = 1.0-4.3), and for men 40 years or younger the ORs were 1.9 (CI = 0.8-4.4) and 3.9 (CI = 1.4-11.2). The results were mainly attributable to non-seminoma, the more malignant type of testicular cancer. Our conclusion is that the results for male breast cancer, based on limited numbers, fail to support the suggested association with MF exposure. The results for testicular cancer gave some support to the hypothesis of a hormonal link between MFs and cancer, and should be further explored.