Background. We examined the association between extremely low-frequency magnetic fields (EMF) and the risk of dementia and Alzheimer's disease using all 9,508 individuals from the Study of Dementia in Swedish Twins (HARMONY) with valid occupational and diagnostic data. Methods. Dementia diagnoses were based on telephone screening followed by in-person clinical workup. Main lifetime occupation was coded according to an established EMF exposure matrix. Covariates were age, gender, education, vascular risk factors, and complexity of work. Based on previous research, data were also analyzed separately for cases with disease onset by age 75 years versus later, men versus women, and those with manual versus nonmanual main occupation. We used generalized estimating equations with the entire sample (to adjust for the inclusion of complete twin pairs) and conditional logistic regression with complete twin pairs only. Results. Level of EMF exposure was not significantly associated with dementia or Alzheimer's disease. However, in stratified analyses, medium and high levels of EMF exposure were associated with increased dementia risk compared with low level in cases with onset by age 75 years (odds ratio: 1.94, 95% confidence interval: 1.07-3.65 for medium, odds ratio: 2.01, 95% confidence interval: 1.10-3.65 for high) and in participants with manual occupations (odds ratio: 1.81, 95% confidence interval: 1.06-3.09 for medium, odds ratio: 1.75, 95% confidence interval: 1.00-3.05 for high). Results with 42 twin pairs discordant for dementia did not reach statistical significance. Conclusions. Occupational EMF exposure appears relevant primarily to dementia with an earlier onset and among former manual workers. © 2010 The Author.