The USA and Europe have maximum contaminant limits (MCLs) of 44 and 50 mg l-1 respectively for nitrate in potable water. Both limits seem to depend on studies relating cases of methaemoglobinaemia in infants to the concentrations of nitrate in the water from associated wells. These studies were made in the mid-west of the USA half a century ago and their limitations were noted at the time. The original investigator commented on the insanitary state of the wells, which were dug too close to barnyards and pit privies, and uncovered so that animal excrement could fall into them. He also noted that bacteria as well as nitrate were present in the well water, but the bacteria, some of which cause gastroenteritis, seem to have been forgotten. In most cases, it was not possible to measure the methaemoglobin level in the blood and the case was defined on the ‘medical suspicion of methaemoglobinaemia'. The structure and treatment of the resulting data-set was not regarded by the American Public Health Association as adequate for defining a safe limit, and the MCL of 44 mg l-1 derived from them seems to have been an artefact. The criteria for inclusion meant there were no cases at nitrate concentrations of 44 mg l-1 or less and only 1.8% of them at concentrations between 45 and 89 mg l-1. Intake of nitrate from water is usually small compared with that from leaf vegetables, and medical research suggests that infant methaemoglobinaemia is frequently caused by gastroenteritis. The misunderstanding about nitrate has led to a world-scale scientific error which has lasted more than 50 years.