The problem posed by electric and magnetic fields (EMF) is an example of a class of problems increasingly faced by environmental epidemiologists. An easily observed characteristic of location is associated with disease. This characteristic is clearly a surrogate for some component of a complex mixture, but there is no compelling biological theory to indicate what the responsible component or components are. Indeed, the lack of theory and a measurable agent leads some to doubt the reality of the association between disease and the surrogate. How does one home in on the responsible component of the mixture? The research strategy for approaching EMF health effects by the California Department of Health Services is described. The author argues, from some preliminary results about the stability of spot measurements and personal monitoring for magnetic field intensity, that the time integral of magnetic field intensity may not be a prime candidate as a causal agent of cancer. An approach is suggested for evaluating locations with cancer clusters in the context of a systematic study which could provide useful information. The approach used for EMF is generalized for other complex mixture problems such as the sick building syndrome. Nine recommendations are made.