In the 1970s, Bangladesh and its development partners set out to defeat one of the country's great killers — disease carried in surface water — by sinking millions of hand-pumped tube wells. The campaign worked against diarrhoea. It also, unknowingly, connected tens of millions of people to water laced with arsenic, a poison occurring naturally in the delta's sediments. Writing in the Bulletin of the World Health Organization in 2000, Allan H. Smith and colleagues called the contamination "the largest poisoning of a population in history". Three decades on, exposure has fallen substantially — but the most recent survey evidence shows millions of Bangladeshis still drinking water above the WHO guideline.
The Tube-Well Revolution
The crisis grew out of an earlier emergency. In the 1970s, the ponds and rivers most Bangladeshis drew water from were heavily contaminated with bacteria, and diarrhoeal diseases — cholera, dysentery, typhoid — were among the leading causes of death, especially among children. Groundwater offered an obvious escape: microbiologically clean, available in almost every village, and reachable with a cheap hand pump.
From the 1970s onwards, UNICEF led an international effort, soon joined by the government, to install hand-pumped tube wells across the countryside, and villagers copied the technology privately on a vast scale. By the time the arsenic problem was recognised, more than four million shallow wells had been installed with support from UNICEF, the World Bank and NGOs alone, according to figures cited in a 2006 House of Lords judgment — and millions more had been sunk privately. Groundwater supplied most rural households, and deaths from diarrhoeal disease had fallen steeply. By the measure it was designed for, the programme was a triumph.
The flaw was in the testing. Standard water-quality checks of the era looked for pathogens and a short list of chemical parameters; arsenic was not among them. As Lord Hoffmann observed in the House of Lords judgment, "at least until 1993, no one thought of testing the tubewell drinking water for arsenic". The water was clear, cool and good-tasting. It was also, in much of the country, quietly toxic.
The Discovery: Chapai Nawabganj, 1993
The warning signs appeared first across the border: in 1983, the physician K. C. Saha identified the first cases of arsenic-induced skin lesions in India's West Bengal, in geologically identical delta sediments (Smith et al., Bulletin of the World Health Organization, 2000). A decade later the same signature was found in Bangladesh. In 1993, the Department of Public Health Engineering (DPHE) confirmed arsenic in tube-well water at Chamagram village in Baroghoria Union, Chapai Nawabganj district, in the far west; in 1994, eight arsenicosis patients were identified in the same village.
Chapai Nawabganj lies within the Barind Tract, the dry upland of the north-west — a reminder that the crisis was never confined to one corner of the delta. The arsenic itself is entirely natural. The great rivers of Bangladesh have carried sediment down from the Himalaya for millennia, and arsenic travels with it, bound to iron oxyhydroxide coatings on the sediment grains. In a landmark Nature paper in 1998, Nickson and colleagues showed that where young delta sediments are buried with organic matter, microbes create oxygen-free conditions in which those iron coatings dissolve — releasing the arsenic into groundwater. The same research explained the pattern that later shaped the response: older Pleistocene aquifers, and deep aquifers below roughly 150 metres, were largely arsenic-free, their sediments flushed or oxidised over far longer timescales.
Recognition lagged behind detection. Through the mid-1990s, university studies accumulated, and in 1998 the British Geological Survey (BGS) and DPHE began the first systematic national investigation. International attention followed — the New York Times published its "Death by Arsenic" investigation in November 1998 — and in 2000 the WHO Bulletin paper by Smith, Lingas and Rahman fixed the phrase by which the crisis is still known.
How Big Was the Crisis?
The definitive early measurement was the National Hydrochemical Survey by BGS and DPHE, published in 2001. Testing 3,534 wells across the country (excluding the Chittagong Hill Tracts), it found that 27 percent of shallow tube wells — those less than 150 metres deep — exceeded Bangladesh's standard of 50 micrograms of arsenic per litre (µg/L), and 46 percent exceeded the WHO guideline of 10 µg/L. Only about 1 percent of deep wells exceeded the national standard. Converting proportions into people, it estimated that 35 million people were drinking water above 50 µg/L and 57 million above 10 µg/L (BGS & DPHE, 2001).
A fourteen-year study by Dipankar Chakraborti's group, analysing 52,202 samples collected from 1996, found similar proportions — 27.2 percent of wells above the national standard, 42.1 percent above the WHO guideline — recorded wells above the national standard in 50 of the 64 districts, and estimated around 22 million people drinking water above 50 µg/L and 36 million above 10 µg/L (Rahman et al., Water Research, 2010). Early assessments cited by Smith and colleagues put the population at risk at 35–77 million.
Two yardsticks explain part of the spread. In 1993 the WHO lowered its guideline value to 10 µg/L. Bangladesh retained its national standard of 50 µg/L — five times higher — and still applies it today. A well can therefore be officially "safe" in Bangladesh while delivering five times the arsenic the WHO advises.
What Arsenic Does to the Body
Arsenic in water has no taste, smell or colour, and chronic poisoning — arsenicosis — works slowly. The first visible signs typically appear only after years of exposure: dark, spotted pigmentation known as melanosis, sometimes mixed with pale patches, followed by keratosis, hard wart-like thickening of the skin on the palms and soles. In advanced cases, poor circulation can lead to gangrene. The skin signs are the visible tip of a systemic poison.
The gravest effects are internal and slower still. Long-term exposure is an established cause of cancers of the skin, lung and bladder, and is linked to cancers of the liver and kidney, to cardiovascular and respiratory disease, and to diabetes. Smith and colleagues estimated in 2000 that as many as one in ten people who drink water containing 500 µg/L over a lifetime may ultimately die of cancers caused by the exposure. The Health Effects of Arsenic Longitudinal Study (HEALS), which followed some 12,000 residents of Araihazar, near Dhaka, for over a decade, found that over a fifth of all deaths in the study population were attributable to water above 10 µg/L, and that those exposed to the highest concentrations (150–864 µg/L) had an all-cause mortality rate nearly 70 percent higher than the least exposed (Argos et al., The Lancet, 2010).
Using 2009 survey data, Flanagan, Johnston and Zheng estimated (Bulletin of the World Health Organization, 2012) that arsenic accounted for roughly 24,000 adult deaths a year among people exposed above 50 µg/L, and perhaps 19,000 more among those exposed at 10–50 µg/L — some 43,000 deaths annually, about one in eighteen adult deaths. Against such figures, the official patient count looks starkly incomplete: the national count reported in the mid-2000s had identified just 38,430 arsenicosis patients, and in 2020 Prothom Alo reported that the health directorate held no updated patient records at all — a gap reflecting the disease's long latency and weak rural diagnosis as much as its true prevalence.
Red and Green: The Great Testing Campaign
The first large-scale response was to find out which water was dangerous. Between 2000 and 2005, under the World Bank-funded Bangladesh Arsenic Mitigation Water Supply Project (BAMWSP, 1998–2006) and partner programmes, field workers tested nearly five million tube wells in arsenic-prone areas with portable kits — 4.94 million by 2003, per figures compiled by Flanagan and colleagues (2012). Each well was painted: red if its water exceeded the national standard, green if within it. By 2005, about 1.4 million wells had been painted red and 3.5 million green.
The message was deliberately simple: drink from a green well, and if your own well is red, share a neighbour's green one. It worked, at first: an expert review cited by Johnston and colleagues (PLoS ONE, 2013) estimated that by 2006, 29 percent of the initially exposed population had switched to safe shallow wells and 12 percent to deep tube wells. But the campaign was a one-off. Paint faded; pumps broke and were replaced by new wells nobody tested. An eighteen-year follow-up in Araihazar found that testing alone had cut exposure sharply in the short term, but most wells in the study area had since been replaced and very few ever retested (Jamil et al., Environmental Science & Technology, 2019). Knowledge, it turned out, decays along with the paint.
Mitigation: What Worked, What Faded
In 2004 the government adopted a National Policy for Arsenic Mitigation and an accompanying Implementation Plan, preferring surface-water and piped supplies where feasible. In practice, the technology that carried the response was the one the 2001 survey had pointed to: the deep tube well. Deep wells were officially a last resort, yet by 2006 more than 100,000 alternative sources had been installed in affected areas and about 70 percent were deep tube wells (Johnston et al., 2013) — cheap, familiar, and drawing on aquifers of which only about 1 percent exceeded the standard (BGS & DPHE, 2001). Their limits are real: they cost more to install, can draw saline water in coastal districts, and depend on a finite deep aquifer that must be protected from leakage down poorly sealed boreholes. In the Barind north-west, deep groundwater is also the irrigation lifeline, and water tables there are falling.
The alternatives each found a niche, and each revealed a weakness. Pond sand filters performed well where a community maintained them, but many fell into disrepair. Dug wells, prioritised in early policy, repeatedly suffered bacteriological contamination; the BAMWSP completion report (World Bank, 2007) judged its emergency dug-well component unsuccessful on water quality. Rainwater harvesting serves mainly the saline coastal belt, where storage limits its use through the dry season. Household arsenic-removal filters had the most loyal users of any option — 92.9 percent regular use among adopters in one national survey (Johnston et al., 2013). The best-known design, the SONO filter developed in 2001 by Bangladeshi chemist Abul Hussam of George Mason University, won the US$1 million Grainger Challenge Prize in 2007, and an estimated one million people were using SONO filters by 2010 (WIPO). Yet household filters never reached more than a small fraction of the exposed population.
Accountability, meanwhile, largely failed. Villagers who sued in Britain over a 1991–92 BGS pilot study of groundwater chemistry — which, following the scientific orthodoxy of the day, had not tested for arsenic — lost at the House of Lords in July 2006, when the court held in Sutradhar v Natural Environment Research Council [2006] UKHL 33 that the survey owed no duty of care to Bangladesh's population. It remains the crisis's leading legal landmark: a catastrophe with millions of victims, and no one legally to blame.
Where Things Stand Today
The trend is genuinely downward — and genuinely unfinished. Successive rounds of the Bangladesh Bureau of Statistics and UNICEF Multiple Indicator Cluster Surveys (MICS) trace the decline. Analysing the 2009 survey, Flanagan and colleagues (2012) estimated about 20 million people exposed above the national standard and 45 million above the WHO guideline. A 2026 analysis of the 2019 MICS data by Pakkos, Alcamo and Shamsudduha (Water International) put the figures at 15.0 million people above 50 µg/L and 24.3 million above 10 µg/L — roughly half the exposure the BGS survey implied at the century's turn, but still a vast population. Human Rights Watch, in a 2016 report, similarly estimated about 20 million people still exposed above the national standard, and criticised the politicised allocation of government deep tube wells.
The newest evidence extends the story to the present decade. A preprint analysis posted in October 2026 (medRxiv; not yet peer-reviewed) of the 2025 MICS round found the share of drinking-water sources above the WHO guideline had fallen to 14.4 percent, from 24.0 percent in 2012–13 — yet that still meant about 26 million people using water above 10 µg/L in 2025, with 8.6 percent of sources above the national standard. Exposure was heavily concentrated: 25 districts high in both 2012–13 and 2019 held 39 percent of the population but 72 percent of those exposed in 2025. Government retesting in 2021–23, cited in the same study, suggested about 20 million people still drank unsafe water.
The deepest problem may now be measurement itself. The blanket screening of the early 2000s has never been repeated; as Prothom Alo reported in 2020, the country went some seventeen years without a comprehensive national survey of its tube wells, and the millions of private wells sunk since the painting campaign are mostly untested. Bangladesh also still judges its water against the 50 µg/L standard rather than the WHO's 10 µg/L — so part of what is officially counted as safe would not be counted as safe elsewhere.
Lessons of a Slow Catastrophe
The arsenic crisis is now a standard case study in public health, its lessons written into water policy worldwide. First, a technology should be tested for the hazards of the place where it is deployed, before it is scaled: the warning existed — West Bengal's cases were identified in 1983, a decade before Bangladesh's first confirmation — and was not connected in time. Second, a one-off campaign is not a system: wells are replaced, paint fades, and only routine monitoring keeps exposure falling. Third, safe water is social as much as chemical — options failed less often on engineering than on distance, maintenance, taste and who could use whose well. Fourth, the deep aquifer, the quiet hero of the response, must be guarded, not mined: the geology that makes deep water safe also makes it slow to renew.
Bangladesh's experience changed what the world means by "safe water": global monitoring under the Sustainable Development Goals now counts chemical quality, including arsenic, alongside access. For Bangladesh itself, the remaining task is narrower and harder — to finish the job in the districts where exposure persists, to test the new wells sunk every year, and to decide whether its standard should match the world's. The largest mass poisoning in history was not an event but a process. Ending it will be one too.