Controversial laboratory results and the now discredited wire code paradox'' suggest that epidemiologists might miss something if they measured only 60 Hz ®elds and summarize time series of exposure with only a time integral. This has led to the question of how else the exposure issue could be addressed. This paper proposes a potential answer to this question by borrowing concepts from pharmacology. The paper brie¯y explores how the limitations of current analysis can lead to poor public policy decisions. The paper then uses a pharmacological analogy to highlight such concepts as mixture,'' ingredient,'' dose duration,'' dosing schedule,'' and windows of biological vulnerability'' to guide an epidemiologist in choosing what to measure. The authors point out that any candidate ingredient'' or exposure metric must be associated with wire code or spot measurements if it is to explain previous epidemiological associations. This helps eliminate many ingredients'' and exposure metrics from consideration. The authors then describe how they applied these concepts in the design of a prospective study of miscarriage and EMF exposure.