Three rules were applied consistently, and they matter for reading the table.

“Last record” means the last evidence the bird was alive — a specimen, a photograph, a sound recording, an observation accepted by the relevant authority, or the death of the last known captive. For the kāmaʻo that is a tape recording of one singing male made on 27 June 1987. For the Kauaʻi ʻōʻō it is David Boynton’s recording of the last male singing a duet with nobody answering.

Where a bird was declared twice, the earlier declaration counts. Eight of these species were removed from the US Endangered Species Act in October 2023; several had already been listed Extinct by the IUCN years before, and those earlier dates are the ones in the table. Using the American dates throughout would have inflated every figure.

The dataset is this site, not the world. Thirteen species is a small sample, weighted toward Hawaiʻi because that is where the extinctions were. It is not a global census, and the median should be read as a description of these birds rather than a published constant.

What survives those caveats is the shape of the distribution. The three short gaps have one thing in common: somebody watched the last bird die. Orange Band, the last dusky seaside sparrow, was found dead on 17 June 1987 at Disney’s Discovery Island, having been caught, banded and housed. The last poʻouli died in captivity on 26 November 2004 with avian malaria. When there is a body and a date, the declaration follows within a few years.

Everywhere else the record simply stops, and the stopping is not an event anybody can put a date on.

Timeline: how the rules got this way

The 1966–67 pair of lines is the oddest thing in the record and it is not a transcription error. The US Fish and Wildlife Service proclaimed the nukupuʻu extinct in 1966, and then listed both species as endangered in 1967. The bird was officially gone and officially protected within twelve months, and the protection outlasted the proclamation by fifty-six years.

Why does it take decades?

Two reasons hold up, and one does not.

The first is that premature declarations are a real error with real costs. Under the US Endangered Species Act, delisting is not a label change. It withdraws the legal protection of habitat. If the bird is still there, the declaration removes the thing keeping it alive. That is why the ivory-billed woodpecker was proposed for delisting in 2021, held back from the 2023 batch, and had its proposal withdrawn entirely in July 2024 — 80 years after the last accepted sighting. The service decided that being wrong in one direction was worse than being slow.

The second is that rediscovery is not rare. Martin and colleagues counted 562 terrestrial vertebrate species that nobody had seen for at least 50 years and that nobody had declared extinct — 38 of them birds. For comparison, 311 terrestrial vertebrates have been declared extinct since 1500. There are 80 per cent more lost species than declared ones, and a meaningful fraction of the lost ones are still out there: we have a whole page of birds that came back, one of them after 330 years.

The reason that does not hold up is that the delay is harmless. Two of the birds in the table above had been gone for more than sixty years when the United States first granted them legal protection in 1967. Both nukupuʻu were last recorded around the turn of the twentieth century. Between 1968 and 1975, searchers spent over 500 days in the field on Kauaʻi looking for them and others, and confirmed nothing. The protection was real, the money was real, and the birds had been dead for two generations.

Meanwhile at least 26 sightings of the Kauaʻi nukupuʻu and 27 of the Maui nukupuʻu were submitted and rejected between 1960 and 1996 — not because the observers were careless, but because a surviving honeycreeper, the ʻamakihi, occasionally grows a deformed bill and looks exactly like a nukupuʻu in a glimpse. Fifty-three reports, no evidence, half a century of maybe.

What happens during the gap?

Not nothing. The decades between a last record and a declaration are filled with searching, and the search effort is where most of the money goes.

The pattern repeats across the species on this site. On Kauaʻi, the 500-plus days of fieldwork between 1968 and 1975 were a deliberate, funded attempt to settle whether the ʻakialoa and the nukupuʻu were still there. On Molokaʻi, surveys for the kākāwahie ran in 1972, 1975, 1979–80, 1988 and 1995, every one negative. Munro searched for a month in 1936 and failed — then noted that he had never reached the upper forests, which is exactly the gap that keeps a species officially alive.

For Jerdon’s courser, which is lost rather than extinct, the modern version of this looks like 200 camera traps and 107,000 images between 2013 and 2015, followed by 17 acoustic recorders in 2019–20. Nothing in either. For the pink-headed duck, a standing cash reward ran from 1930 to 1960 and was never claimed, and BirdLife surveyed Kachin State in Myanmar between 2003 and 2005 on the strength of unconfirmed reports.

Searching is how the doubt gets removed, and removing doubt from a negative claim is slow, expensive work that produces nothing publishable when it succeeds.

Why does the same bird have two extinction dates?

Because two authorities decide independently, and they do not synchronise.

The Kauaʻi ʻōʻō is the clearest case. The IUCN listed it Extinct in 2000. The United States removed it from the Endangered Species Act on 16 November 2023 — twenty-three years later, for the same bird, on the same evidence. For the poʻouli the two dates are 2019 and November 2023, four years apart. For the Carolina parakeet, the IUCN’s extinction date is 1920 while the American Ornithologists’ Union did not accept extinction until 1939, and the last captive bird had actually died in February 1918 — three dates, none of them the same.

The paradise parrot shows the cost of the lag in a single species. Australia’s only extinct mainland bird was last reported around Gayndah in 1928. The IUCN first listed it as Threatened in 1988 — sixty years after the last record — and only reclassified it as Extinct in 1994. For six years it was formally a species in trouble rather than a species that was gone.

So a reported extinction date is always a date on a decision, not a date on a death. Which authority made it, and when they happened to meet, changes the figure by decades.

Which birds have never been declared extinct at all?

These four are on this site, all of them with a last record in living memory or close to it, none of them formally extinct.

“Possibly Extinct” is the Red List’s honest shrug, and it is doing a lot of work. All four of these birds are, on the evidence, almost certainly gone. The last imperial woodpecker anyone can verify is a female in 85 seconds of 16 mm film shot by William Rhein in Durango in 1956; when researchers found the filming site again in 2010, the forest had been logged. The last pink-headed duck accepted in the wild was trapped in Bihar in June 1935, and a standing cash reward for physical evidence went unclaimed for thirty years.

None of that is the same as proof, so the category stays.

When did these birds actually die?

Not when the records stopped, and not when the paperwork caught up. Somewhere in between, and much closer to the first.

There is a body of work on exactly this. Elphick, Roberts and Reed modelled extinction dates for North American and Hawaiian birds from their sighting records in 2010; Roberts and Jarić extended the method in 2016 to handle sightings of uncertain reliability. The outputs are windows rather than dates, and the windows sit close behind the last record:

• Hawaiʻi mamo — last seen July 1898, modelled extinction 1911–1915
• Kauaʻi ʻakialoa — last seen April 1969, modelled extinction 1967–1973
• Kāmaʻo — last recorded June 1987, modelled extinction 1991–1999
• Both nukupuʻu — modelled extinction about 1901, upper limit 1906
• Labrador duck — last credible sighting 1878, estimated extinction 1880

Set those against the declaration dates and the arithmetic is bleak. The model says the Kauaʻi nukupuʻu was gone by 1906 at the outside. The United States said so in 2023. On the best available estimate of when the bird actually died, the paperwork trailed it by 117 years.

The kākāwahie is the exception that shows the limits of the method: its modelled window runs 1910 to 1969, which is no window at all. A bird collected 36 times in three months in 1907 and then found twice by one observer in the early 1960s produces a sighting record the models cannot resolve.

Both dates are now in one place for every species on this site. The extinct birds explorer plots all thirty-one on a map, a timeline and a sortable table, naming for each one what the last record actually was — a specimen, a sighting, a photograph, a sound recording or a death in captivity. Fourteen of the thirty-one end with a sighting, which is the weakest evidence on the list and a large part of why the declarations take so long.

Interesting facts

• The shortest gap on this site, three years, belongs to a bird that was captured, banded, given a name and housed at a Disney theme park. Precision about extinction requires custody.
• The USFWS proclaimed the nukupuʻu extinct in 1966 and listed it as endangered in 1967. Both statements stood simultaneously for 56 years.
• The slender-billed curlew’s last irrefutable record and its extinction declaration are 30 years and 7 months apart — 25 February 1995 at Merja Zerga in Morocco, 10 October 2025 in Bonn.
• 38 bird species worldwide have gone unseen for over 50 years without being declared extinct.
• The last evidence for two species on this list is audio, not sight: a tape of one kāmaʻo singing in 1987, and a recording of the last Kauaʻi ʻōʻō performing a duet alone.

FAQ

How long does it take to declare a bird extinct? On the thirteen documented cases on this site, a median of 36 years after the last confirmed record, and a mean of 47.5. The range is wide: three years when the last bird died in custody, 124 years when the record simply stopped.

Is there a 50-year rule for declaring extinction? No. A 50-year no-sighting threshold was proposed in the 1980s and abandoned, because too many species were rediscovered after longer gaps than that. The IUCN instead requires “no reasonable doubt that the last individual has died”, judged case by case.

Why are some birds listed as “Possibly Extinct”? Because the evidence points strongly to extinction but falls short of the IUCN’s standard. Four birds on this site sit in that category, with last records from 1935 to 1963. It is a statement about the evidence, not a prediction that the bird will turn up.

How many birds are “lost” but not declared extinct? 38, in the global count published by Martin and colleagues — species unseen for at least 50 years with no formal declaration. Across all terrestrial vertebrates the figure is 562, against 311 declared extinct since 1500.

Which bird waited longest to be declared extinct? Of the birds covered here, the Kauaʻi nukupuʻu: last collected 6 May 1899, delisted 16 October 2023. 124 years.

Does declaring a species extinct do any harm? It can. Under the US Endangered Species Act, delisting removes habitat protection, so a premature declaration withdraws the safeguard from a bird that may still be alive. That risk is why the ivory-billed woodpecker’s delisting was proposed in 2021 and formally withdrawn in July 2024.

Who actually makes the declaration? Usually the IUCN, through its Red List assessments for BirdLife International, or a national authority such as the US Fish and Wildlife Service under the Endangered Species Act. The two work on separate timetables and often disagree by decades.

Sources

• US Fish and Wildlife Service — 21 species delisted from the Endangered Species Act due to extinction (16 October 2023), the primary announcement
• American Bird Conservancy — USFWS extinction declarations 2023
• Martin, T.E., Bennett, G.C., Fairbairn, A. & Mooers, A.O. (2022). ‘Lost’ taxa and their conservation implications. Animal Conservation 26(1): 14–24 — the 562 lost vertebrates and 38 lost birds
• Simon Fraser University — Lost or extinct? Study finds the existence of 562 animal species remains uncertain (19 May 2022)
• Mooers, A., Martin, T. & Bennett, G. — When is a species really extinct?, The Conversation, 23 May 2022, on the abandoned 50-year rule
• Elphick, C.S., Roberts, D.L. & Reed, J.M. (2010). Estimated dates of recent extinctions for North American and Hawaiian birds. Biological Conservation 143: 617–624
• Roberts, D.L. & Jarić, I. (2016). Inferring extinction in North American and Hawaiian birds in the presence of sighting uncertainty. PeerJ 4: e2426
• UNEP/AEWA — Slender-billed Curlew officially declared extinct (10 October 2025)
• Per-species dates are sourced on the individual pages linked in the tables above. The dataset and the arithmetic behind the median are kept in content/supplements/declaration-lag.json.

Image credits

• Featured image: Carolina parakeet museum mount, by stewartbrandphotos, via Wikimedia Commons, CC0.
• Kauaʻi ʻōʻō mounted specimen, Bishop Museum, Honolulu, by Hiart, via Wikimedia Commons, CC0.
• Photograph of a living dusky seaside sparrow, by P. W. Sykes, US Fish and Wildlife Service, via Wikimedia Commons, public domain.
• Imperial woodpecker male specimen RMNH.AVES.110098, Naturalis Biodiversity Center, via Wikimedia Commons, CC BY-SA 3.0.
• Eskimo curlew, Redpath Museum, McGill University, by Daderot, via Wikimedia Commons, CC0.