In this study, we introduced and tested a new approach to characterize residential magnetic field (MF) exposure. Short-term 20-min MF measurements were obtained by a person who carried out instantaneous spot measurements in residences. Compared to spot measurements, the 20-min measurement could potentially improve exposure assessment, because it contains information of temporal variations of MF, which have been suggested as biologically important characteristics of MF exposure. We have used this new exposure assessment method on a study of maternal MF exposure and reproductive outcomes. To validate the new method, the exposure of 30 subjects was measured with a more accurate "gold standard" method (24 h personal exposure measurements). The measures of validity used were the Spearman correlation coefficient (r), sensitivity, and specificity. We evaluated the validity of the 20-min measurements for estimating several different exposure metrics for the entire 24 h measurement period or for the hours spent at home: arithmetic mean, median, percentage of time above 0.15 microT, standard deviation, rate of change metric, standardized rate of change metric, constant field metric, and three metrics for the occurrence of high-peak exposures. The 20-min measurement was modestly associated with standard deviation and the rate of change metric, but gave very little information of other metrics of temporal variation. The 20-min measurement can also be used for assessing exposure metrics such as arithmetic mean and median, but it does not seem to offer any advantages compared to traditional 'spot' measurements. The 20-min measurement was not useful for assessing occurrence of high-peak exposures. We conclude that the 20-min measurement is useful for estimating some aspects of MF temporal variability.