Stand inside one of Odisha’s concrete cyclone shelters today, on stilts above the flood line, built to hold a thousand people and their livestock, and it is hard to picture what the coast looked like before these buildings existed. I went looking for that before-and-after because the story of Odisha cyclone evacuation is not a slogan, it is a twenty-five-year engineering and governance project that turned India’s worst disaster story into its best evacuation record. It started from the single worst act of nature the state has ever recorded.
Odisha Cyclone Evacuation: Where the Story Begins
On the night of October 29, 1999, a super cyclone with wind speeds exceeding 250 kilometres an hour made landfall near Paradip on Odisha’s coast, then stalled over the state for nearly 30 hours, dumping torrential rain on top of a storm surge that in places reportedly reached six metres. According to the Odisha State Disaster Management Authority, the official death toll stood at 9,887. A white paper prepared by the state government the following year put the figure at a similarly staggering 9,885, while independent estimates at the time ran considerably higher. Entire coastal villages in Jagatsinghpur and Kendrapara districts were erased. Livestock losses ran into the millions. It remains the deadliest cyclone to strike India in living memory.
In the aftermath, Odisha did something no other Indian state had done: it stood up its own dedicated disaster management authority. The Odisha State Disaster Management Authority, OSDMA, was created in 1999, six years before the National Disaster Management Authority existed at the federal level. That sequencing matters. Odisha was not implementing a national mandate handed down from Delhi. It was building disaster governance from a wound, ahead of the country that would eventually copy parts of its model. It is worth sitting with how unusual that was for the period: in 1999, disaster response in most of India still meant relief camps assembled after the fact, not infrastructure and drills built years in advance of a storm nobody could yet name.
Odisha did not wait for a national disaster authority to tell it what to build. It built one first, out of the wreckage of 1999.
Building for the Next Storm
OSDMA’s early work was unglamorous and physical: constructing multipurpose cyclone shelters along the coast, most of them doubling as schools or community halls in ordinary weather. Funding for a large share of that shelter network came through the National Cyclone Risk Mitigation Project, backed in part by World Bank financing, which paired shelter construction with early-warning infrastructure, evacuation route planning, and coastal embankments. Odisha also set an explicit policy goal that would have sounded unrealistic in 1999: zero casualties from cyclones, achieved not by stopping the storms but by making sure nobody was left in their path.
- 1999: OSDMA established; roughly 9,900 deaths from the super cyclone.
- 2013, Cyclone Phailin: Around one million people evacuated in about 36 hours; official death toll in the dozens.
- 2019, Cyclone Fani: More than a million people moved into over 4,800 shelters; death toll again in the dozens, a fraction of 1999’s toll.
Phailin: The First Real Test
Cyclone Phailin arrived in October 2013, roughly as powerful at landfall as the 1999 storm by wind speed, making landfall near Gopalpur. This time, Odisha’s administration evacuated close to a million people, most of them within about 36 hours, moving entire coastal communities inland or into the cyclone shelters that OSDMA had spent a decade building. The World Bank later cited the Phailin response as a case study in disaster preparedness paying for itself. Official death figures for Phailin in Odisha came in at under fifty, compared to nearly ten thousand fourteen years earlier, from a storm of comparable strength. That is not a small improvement. It is close to a two-order-of-magnitude change in outcome, achieved primarily through evacuation logistics and shelters rather than any change in the storm itself.
Fani and International Recognition
By the time Cyclone Fani struck in May 2019, Odisha’s evacuation machinery was more practised still. More than a million people, by most accounts around 1.2 million, were moved out of the storm’s path and into shelters, with reports counting close to five thousand shelters activated across the coast. Officials leaned on India Meteorological Department forecasting, door-to-door warnings delivered by local volunteers, and a state government that had rehearsed this exact drill twice before. United Nations agencies publicly praised the response in its aftermath, holding it up as a model for disaster risk reduction elsewhere in the world. Odisha’s death toll from Fani, while real and not something to minimise, again came in at a small fraction of 1999’s toll, in the dozens rather than the thousands.
None of this means Fani was painless. Reporting on the aftermath documented widespread damage to homes, crops, and fishing livelihoods along the coast, and it takes more than a successful evacuation to rebuild a fishing family’s boat or a farmer’s flattened paddy field. Odisha’s model has consistently been better at preventing death than at preventing loss. Those are different problems, and only one of them has been solved.
How Odisha Compares to Its Neighbours
Odisha did not build this playbook in a vacuum, and it is not the only place around the Bay of Bengal that has had to learn these lessons the hard way. Bangladesh’s Cyclone Preparedness Programme is often cited internationally as a model Odisha’s own system draws on. It is a volunteer-driven early-warning and evacuation network, built after the catastrophic 1970 Bhola cyclone, which killed an estimated several hundred thousand people. Its core ingredients: village-level volunteers, colour-coded warning signals, and pre-positioned shelters. West Bengal, Odisha’s neighbour to the north, ran its own large-scale evacuation for Cyclone Amphan in May 2020. It moved roughly 300,000 people into shelters, part of a regional evacuation effort of close to three million people across eastern India and Bangladesh. West Bengal did this while simultaneously managing early Covid-19 restrictions, a logistical complication Odisha never had to contend with during Phailin or Fani.
What sets Odisha apart within India specifically is not that it evacuates people, most cyclone-prone coastal states now do, but that it built the institution first and kept it running for a quarter century without waiting for a national framework to arrive. It is the same kind of early-mover pattern this series found in how Tamil Nadu built India’s best social outcomes: a state government treating a hard problem as a decades-long institutional project rather than a one-term political promise. Andhra Pradesh, hit by many of the same storms that cross Odisha’s coast, has since built out its own shelter network partly modelled on OSDMA’s approach. The lesson seems to have travelled along the coastline even where the exact death-toll numbers have not been matched.
A Coastline Under More Pressure, Not Less
None of this is happening against a stable backdrop. Climate scientists tracking the Bay of Bengal have flagged warming sea-surface temperatures as a factor that can intensify cyclones more rapidly in the days before landfall, giving forecasters and evacuation planners a narrower window to act. Odisha’s coast is also low-lying and densely populated in its cyclone-exposed districts, which means the population that needs to be moved on short notice keeps growing even as the state’s shelter and evacuation infrastructure matures. Sustaining a zero-casualty policy under those conditions is not a one-time achievement Odisha can bank and move on from. It is a target that gets harder to hit with every passing decade, not easier, which is exactly why the institutional continuity behind OSDMA matters as much as any single storm response.
What Actually Changed
Talk to disaster management researchers who have studied the Odisha model and a few components come up repeatedly, none of them exotic. Better and more localised forecasting from the India Meteorological Department, communicated through a chain that reaches village level rather than stopping at the district headquarters. A standing shelter network, physically pre-positioned, rather than improvised evacuation centres set up after a warning is issued. Trained volunteer networks who go door to door rather than relying on people to see a television warning or hear a siren. And political continuity: successive state governments in Odisha have kept disaster preparedness as a functioning bureaucratic priority between cyclones, not just during them, which is the part hardest for other disaster-prone states to replicate. It is the same continuity that is often missing elsewhere in Indian local governance, where, as this publication has reported, panchayats raise barely a fraction of their own money and struggle to fund sustained, unglamorous institutional work between crises.
That continuity is worth dwelling on, because it is the least visible part of the story and probably the most important one. Shelters decay if nobody maintains them between storms. Volunteer networks dissolve if nobody keeps training new members. Early-warning systems atrophy if the same drill is not run and rerun in the years when no cyclone comes. Odisha’s achievement was not a single well-executed evacuation. It was building an institution, OSDMA, that treated the gap years between disasters as part of the job, not a pause in it.
The Model, and Its Limits
Odisha’s evacuation record has become something other coastal states and disaster planners elsewhere hold up as a reference point, and there is genuine reason for that. Getting a million or more people out of a storm’s path in under two days, repeatedly, across three major cyclones spanning two decades, is a level of operational competence that few disaster-prone regions anywhere in the world can claim. But the model built to save lives was never designed to save livelihoods, and every cyclone since 1999 has left behind flattened paddy fields, destroyed fishing boats, and damaged homes that a shelter cannot protect. Odisha solved the part of the disaster-response problem that kills people quickly. It has not yet solved the part that impoverishes people slowly, in the months after the wind dies down and the shelters empty out.
Standing in that empty cyclone shelter, built for an emergency that might not come again for years, it struck me that Odisha’s real achievement is patience: the willingness to keep funding, staffing, and drilling a system for a disaster nobody can predict the exact date of. Whether that patience extends to rebuilding what the storms take, once the evacuation buses have gone home, is the next chapter Odisha has not fully written yet.