The assumption that people using the same water supply have similar intakes of minerals from drinking water is examined. Duplicate samples of all water drunk during a 24 hour period, including that boiled for beverage preparation, were collected by 109 adult subjects. The results showed that there may be more than a tenfold variation in the amount of water people drink daily. In addition it was found that the mineral concentrations in the 24 hour samples can vary markedly from those in water collected from the source or supply or from household taps which are the usual sampling points for epidemiological studies. The convention of using total hardness as an indicator of mineral intakes from hard and soft water is questioned.