Water

Dispensers for Safe Water

Research summary, last reviewed May 2026. US 501(c)(3), EIN 90-0874591 (Evidence Action, Inc.). Official site.

Independent recognition: GiveWell grantee (grant of up to US$64.7m, 2022) · Giving What We Can recommended · The Life You Can Save recommended (Evidence Action) · Open Philanthropy funded · Charity Navigator 4/4 stars (Evidence Action)

Headline outcome: Around 2.8 million people reached with chlorinated drinking water, at under US$1.50 per person per year

What does Dispensers for Safe Water do?

Dispensers for Safe Water is a programme of Evidence Action, a US nonprofit that scales interventions with strong research behind them. It installs bright blue chlorine dispensers directly beside rural water sources: springs, shallow wells and other collection points. A community member fills their jerry can, turns a calibrated valve to add a measured dose of dilute chlorine, and the water disinfects on the walk home and during household storage. The chlorine is free at the point of use, and a community-elected promoter, most often a woman, keeps the dispenser stocked and teaches neighbours how to use it.

The model began as a 20-dispenser pilot in western Kenya in 2007, built on randomised trials led by Michael Kremer, who later won the 2019 Nobel Prize in Economics. After the 2025 programme transition, Evidence Action operates roughly 18,000 dispensers across Kenya and Uganda, reaching an estimated 2.8 million people, alongside a growing in-line chlorination programme that treats piped water automatically with no behaviour change required.

Why did EveryDrop choose it?

Dispensers for Safe Water passed our four-stage vetting assessment: a clearly aligned charitable purpose (advancement of health and saving of lives), no evidence of undue private benefit, no evidence of public disbenefit, and no undue restrictions on who benefits. Anyone who collects water at a treated source can use the dispenser, free, with no eligibility test of any kind.

It also has an unusual quality we value highly: a track record of following its own evidence even when the news is bad. When independent surveys in 2024 found that fewer people were using the dispensers than internal monitoring had suggested, Evidence Action published the finding, accepted a lower cost-effectiveness estimate, and deliberately scaled the programme back to its highest-impact areas. That is the behaviour the rest of the sector should aspire to.

How effective is it?

The underlying science is among the strongest in global health. A 2023 meta-analysis by Kremer and colleagues, published in the Quarterly Journal of Economics and covering 52 randomised trials, found that water treatment interventions reduce the odds of all-cause under-five mortality by roughly 25 to 28%, with point-of-collection chlorine dispensers costing an estimated US$27 to US$66 per expected DALY averted, among the most cost-effective health interventions ever measured.

GiveWell, the programme’s principal independent evaluator, currently assesses the 2022 grant at around 5 times the cost-effectiveness of unconditional cash transfers (revised down from 7 times). Delivery costs Evidence Action less than US$1.50 per person per year; users pay nothing.

What are the open questions?

We publish what we are monitoring as well as what we like. The big event here is the 2024 to 2025 coverage revision: independent surveys found roughly 40% fewer people using dispensers in Kenya than monitoring had implied, and roughly 70% fewer households with detectable chlorine in their stored water in Uganda and Malawi. GiveWell did not renew the Kenya portion of its grant, published a formal lookback in September 2025, and Evidence Action is winding down dispensers in Malawi entirely while narrowing Kenya and Uganda.

Importantly, what was revised is the realised coverage of one delivery model, not whether chlorination works. The population still reached is still protected, and the newer in-line chlorination model removes the behaviour-change step that caused the shortfall. We deliberately avoid the older “5 million people reached” framings, and we expect Evidence Action to publish restated cumulative impact figures, which we will adopt. We re-check the evidence at every annual review.

Sources

This summary is derived from EveryDrop's internal vetting dossier, prepared under our four-stage assessment framework using public, verifiable sources. Assessments are re-checked every twelve months. If you spot something we should know, email [email protected].