For countries such as Bangladesh with a significant groundwater arsenic problem, there is an urgent need for the arsenic-contaminated wells to be identified as soon as possible and for appropriate action to be taken. This will involve the testing of a large number of wells, potentially up to 11 million in Bangladesh alone. Field-test kits offer the only practical way forward in the timescale required. The classic field method for detecting arsenic (the 'Gutzeit' method) is based on the reaction of arsine gas with mercuric bromide and remains the best practical approach. It can in principle achieve a detection limit of about 10 mug l(-1) by visual comparison of the coloured stain against a colour calibration chart. A more objective result can be achieved when the colour is measured by an electronic instrument. Attention has to be paid to interferences mainly from hydrogen sulfide. Due to analytical errors, both from the field-test kits and from laboratory analysis, some misclassification of wells is inevitable, even under ideal conditions. The extent of misclassification depends on the magnitude of the errors of analysis and the frequency distribution of arsenic observed, but is in principle predictable before an extensive survey is undertaken. For a country with an arsenic distribution similar to that of Bangladesh, providing care is taken to avoid sources of bias during testing, modern field-test kits should be able to reduce this misclassification to under 5% overall.