The main objective of this article is to present a new model that can be used to estimate the exposure of a population to lead contamination from the drinking water supply. The model is not to designed predict the particular situation in any one individual household, but to provide an estimate of the average daily quantity of lead ingested by a consumer population from water quality and the characteristics of the household plumbing system. Data were gathered from field studies carried out over a one-week period. The sites selected were known to have differing risks of contamination due to lead piping. Water supplies were grouped into three classes (low, average and high risk of contamination) as a function of the water's lead dissolving capacity. The first class concerned water of low alkalinity (3.4-6.6°f) and high pH (>7.8), the second, water of high alkalinity (20-25°f) and moderate pH (7.3- 7.8), and the third water of low alkalinity (6°f) and pH close to neutral (7.2). Lead concentrations measured in samples of water having stood in lead piping showed that the rate of corrosion varied exponentially with time, being comparatively high at first. This may be related to the mass of lead released from the internal surface of the lead pipes. Measured levels fell into groups in accordance with the defined classes of water. Statistical techniques were used to calculate coefficients with the aim of characterising the time course of lead flux per unit surface area for each water class more precisely. In flowing water, surface flux at any instant principally depends on the alkalinity and pH of the water. However, the experimental data showed that the internal diameter of the pipe along with temperature are equally important factors. An estimate of the rate of lead dissolution from the pipe surface may be produced by statistical analysis of these four parameters. The daily volume of water used for cooking and drinking purposes at each site was estimated using specially designed 'proportional taps'. The sample of water drawn off, representing 5% of total water consumption, was analysed and its lead concentration determined. The lead dissolved in this water stemmed partly from standing water and partly from flowing water. Further measures of lead concentration were taken to determine the degree of contamination occurring in standing and in flowing water. The comparison of these concentrations with the concentration found within the water sample drawn off gave the percentage of contamination to be attributed to the standing and flowing water fractions. The proportion of total water subject to stagnation was determined from the volume of the plumbing system. The data obtained were used to develop a model and a method for estimating lead contamination of cooking/drinking water and the related risk to the consumer. The proposed approach is based on the following simple observation. Most drinking water supply systems are composed of lead and nonlead piping. Thus, when water is run from a tap, the first fraction of water collected carries the highest lead concentration, since this water was standing in the lead pipes. The following water volume, held in the nonlead piping, will only be contaminated during its flow through the lead piping. The method involves two stages. Firstly, the pH and alkalinity of the water, as well as the lead piping characteristics in the household plumbing system and the stagnation time are used to determine the lead concentration in standing and flowing water. Secondly, using the previously acquired knowledge of the daily consumption of cooking/drinking water for each household, a hydrogramme based on standard consumer behaviour can be established incorporating volumes of cooking/drinking water associated with stagnation times. By combining this information with contamination rates established in the first part of the study, the mean daily concentration of lead in the drinking/cooking water can be calculated. The quantity of lead ingested daily was calculated from a standard daily consumption of 3 l divided up into three equal draw off volumes corresponding to stagnation periods of 8, 3 and 5 h for the morning, midday and evening, respectively. In order to determine the respective concentrations of lead in water standing in and flow through the pipes, the relationships established in the first part of this study (relating to the unit area flow of lead during each of the two phases) were used. The mean daily concentration of lead in the cooking/drinking water was determined by combining these values, taking into account the proportion of water contaminated during each of the two phases. This value could be estimated by reference to the capacity of the lead pipe. In order to improve our knowledge of how different factors, such as water composition and piping characteristics, affect the mean lead concentration in water ingested by consumers, we carried out a computer simulation for lead pipes measuring 20 and 30 mm in diameter and 5 to 40 m long. For water classified as having a high risk of dissolution (pH 7.2 and alkalinity 5°f), the simulation showed that the limit for lead concentration recommended by the WHO for drinking water (10 μg l-1) is reached for five metres of lead piping. Lengths of piping of this order are frequently encountered in old housing. For water at pH 8 and alkalinity 5°f (the low risk class), the recommended limit is reached for pipes measuring 10 m and longer. An average concentration of 50 μg l-1 is reached for high risk waters (pH < 7.2, alkalinity 5°f) only when pipe length was at least 40 m. (C) 2000 Published by Elsevier Science Ltd