The accurate and valid measurement of personal exposure to magnetic fields poses a major challenge for epidemiologic studies. When considering the various methods to assess exposure, it is unclear which measures are most relevant for studies of human disease, if any. Given these uncertainties, the Electromagnetic Fields and Breast Cancer on Long Island Study (EBCLIS) undertook a pilot study to develop the data collection protocol for a case-control study of breast cancer and magnetic fields. The pilot study used and compared various methods to assess residential exposures to magnetic fields, and related these measures to personal exposures. It included 31 women without breast cancer (mean age, 63 AE 7 yr) who lived in their present homes for at least 15 yr. The pilot study consisted of an in-home interview, spot and 24-h magnetic field waveforms and broadband recordings, ground currents, wire coding, and personal 24-h broadband measurements. From the regression analyses, the model that best predicted personal magnetic field exposures included 24-h measurements in the bedroom and in the most lived-in room; as well as ground current test loads taken at the center of this most lived in room (r 2 86%). The addition of other variables in this regression model yielded only small and nonsignificant increases in r 2 . As a direct result of this pilot, EBCLIS will include ground current measurements in its protocol, which have not previously been collected as part of an epidemiologic study. Ground currents may be important because they may be richer in 180 Hz components than are the other currents in a power system. EBCLIS will have the opportunity to examine the ground-current hypothesis in the context of female breast cancer. Bioelectromagnetics 20:487±496, 1999.