The variation in lead concentration was measured by thermal ionisation mass spectrometry isotope dilution in household tap water throughout the day when the plumbing system was not fully flushed. After collection of an initial 125-ml water sample containing 119 μg/l and a 2-1 sample, 125-ml samples were collected at hourly intervals for 8 h. The concentrations in the hourly samples remained in the range 35-52 μg/l compared with 1.7 μg/l for fully flushed water. High precision lead isotopic measurements showed that approximately 50% of the lead in these water samples derives from the tap 'housing' compared with the overall household system. A health risk assessment was performed employing the US Environmental Protection Agency Integrated Exposure Uptake Biokinetic Model. Predicted blood lead levels in infants only exceeded the 'levels of concern' of 10 μg/dl when 100% of the water consumed contained 100 μg Pb/l. It would appear that unless the infant consumed 100% of first flush water at lead concentrations of approximately 100 μg/l, the blood lead would not exceed the recommended 'level of concern'. However, if more than 0.5 l was consumed in drinks and formulae using first flush water, then the blood lead could easily exceed the recommended level. Likewise, a pregnant mother could be at risk of consuming considerably more than the 0.5 l/day first flush water of the concentrations measured, or throughout the day, if the system were not fully flushed.