The objective of this study was to provide evidence to evaluate the proposed National Children's Study (NCS) protocol for household water sampling in rural study areas. Day-to-day variability in total trihalomethane (TTHM) concentrations in community water supplies (CWS) in rural areas was determined, and the correlation between TTHM concentrations from household taps and CWS monitoring reports was evaluated. Daily water samples were collected from 7 households serviced by 7 different CWS for 15 days. Coefficients of variation for TTHM concentration over 15 days ranged from 8% to 20% depending on the household. Correlations were tested between TTHM household concentrations and the closest date- and location-matched CWS monitoring reports for the 15-day mean (R=0.85, P<0.01). To simulate the NCS-proposed protocol, correlations were tested for 30 additional NCS household samples (polynomial fit: R=0.74, P=0.04). CWS reported TTHM concentrations >50 μg/l corresponded to measured NCS household concentrations ranging from 2 to 60 μg/l. TTHM concentrations were higher in CWS than NCS samples (11.2±3.2 μg/l, mean difference±SE, P<0.01). These results show that in rural areas there is high variability within households and poor correlation at higher concentrations, suggesting that TTHM concentrations from CWS monitoring reports are not an accurate measure of exposure in the household.