The Childhood Leukemia Survival Study is examining the possible association between magnetic field exposure and survival of children with newly diagnosed acute lymphocytic leukemia (ALL). We report the results of serial 24-h personal magnetic field monitoring for 412 US and Canadian children and present the correlations between annual values. The mean time-weighted average (TWA) and geometric mean (GM) were similar for first, second, and third year measurements [TWA: 0.11 microT (n = 412), 0.13 microT (n = 304), and 0.12 microT (n = 134), respectively]. There were no consistent differences in mean TWA or GM based on age or gender. Significantly lower mean TWA and GM were found for children living in rural areas. Higher exposures were noted among children living in urban areas, among apartments dwellers, and those living in rental homes. Measurements taken during summer months and among children residing in the northeast and Canada also tended to be higher. Correlations for most metrics were increased among children who had annual measurements performed during the same season. The metric with the highest year-to-year correlation was the GM. The lowest correlations were found for metrics estimating field intermittency and temporal stability. First to second year GMs were well correlated when taken in the same home (Spearman rank correlation = 0.70), but a lower correlation (0.44) was noted among residentially mobile children. Our findings suggest that summarizing exposure using a single measurement of GM can estimate exposures for residentially stable children, but is not a good predictor of personal exposures among children who change residence during the study interval.