December 3, 1984

Bhopal disaster

Runaway reaction releases a lethal gas cloud over Bhopal

The system meant to burn escaping gas had been out of action for months. The tank's refrigeration coolant had been removed. Most of Bhopal was asleep.

About 1 a.m., 3 December 1984

Late on 2 December 1984, workers at Union Carbide India Limited's pesticide plant detected rising pressure and a small methyl-isocyanate leak. At about 1 a.m. on 3 December a safety valve gave way and discharged a toxic plume over nearby Bhopal neighborhoods.

Water had mixed with roughly 40 tons of MIC in storage tank E610, producing a runaway exothermic reaction. The vent-gas scrubber had been turned off, the refrigeration coolant removed and the flare system was out of action. How the water entered E610 remains contested.

Streets and hospitals

The plume struck densely populated settlements while residents slept. Streets filled with human and animal bodies, and local hospitals were rapidly overwhelmed without adequate information about the chemicals involved.

A scholarly review estimates at least 3,800 immediate deaths, with published estimates reaching up to 10,000 in the first few days. The Indian government reported more than 500,000 people exposed.

Not simply an MIC leak

The plant was operated by Union Carbide India Limited, an Indian subsidiary majority-owned by the US-based Union Carbide Corporation — neither simply an American plant nor simply an Indian one.

And what came out of E610 was not simply MIC. Later chemical analysis reported more than 28 substances or reaction products connected with the tank, the residue of heating and reaction inside it. Doctors that night were treating thousands of choking patients without reliable information about what they had inhaled, and the precise composition of the cloud is disputed to this day.

Sources

Researched 23 Aug 2026 4 sources 2 audit passes, 1 issue open

How this was checked

Researched from the web into a fact sheet, rewritten from that sheet, then audited against it by a different model. How the pipeline works →

What the sources leave uncertain

The disaster began late on December 2 and reached its catastrophic release phase around 1 a.m. on December 3; both “night of December 2–3” and December 3 are accurate formulations.
Death totals depend heavily on the time window and definition. Immediate, first-days, officially compensated and modeled long-term totals must not be merged.
The precise mixture in the toxic cloud and the exact path by which water entered E610 remain disputed.