Every orbital launch attempt since Sputnik, 7,416 of them, from Jonathan McDowell's catalogue of everything humans have put into space. Where rockets leave from, who flew them, how often they failed, and where the hardware comes back down: 34,974 catalogued objects so far, with the biggest steered towards Point Nemo, the loneliest spot in the ocean.
On 4 October 1957 a Soviet R-7 rocket put a polished metal ball called Sputnik into orbit, and the space age began. Since then rockets have tried to reach orbit 7,416 times. Jonathan McDowell, an astronomer at the Harvard and Smithsonian Center for Astrophysics, has spent decades cataloguing every one of them, along with every satellite, rocket stage and fragment they left behind. His General Catalog of Artificial Space Objects (GCAT) is free for anyone to use, and it is the source for everything on this page.
The map below shows where those launches left from. Press play and watch the space race spread out from Kazakhstan, northern Russia and Florida, stall after the Soviet Union fell, and then explode in the 2020s as SpaceX's reusable Falcon 9 began flying every two or three days. Then scroll on to see what comes back: about half of everything ever catalogued has already fallen out of the sky, mostly unnoticed, and a few of the largest pieces were steered on purpose into a patch of the South Pacific known as the spacecraft cemetery.
Circles: each spaceport, with area proportional to its orbital launch attempts in the chosen years, split by who launched them. Dark rings: failures, on the same scale. Sites that sit next to each other are merged (Cape Canaveral with the Kennedy Space Center, Baikonur's old and new codes); air launches are placed at the airfield they took off from. Small dots (Re-entry places in Map layers): 2,136 objects whose place of re-entry or landing GCAT records. Navy circle: Point Nemo and an indicative 2,500 km circle around it, not an official boundary. Source: GCAT (J. McDowell, planet4589.org/space/gcat), CC BY 4.0. Outlines: Natural Earth. Base map tiles © Esri.
In the first full year of the space age, 1958, there were 28 attempts and 20 of them failed. By the mid 1960s the United States and the Soviet Union were each launching more than once a week, and the Cold War peak came in 1967, with 139 attempts worldwide. The American total peaked a year earlier at 78; the Soviet one kept climbing, to 108 in 1982, much of it from Plesetsk in the far north, which peaked at 70 launches in 1977.
Orbital launch attempts per year, by who launched them
Stacked by the country of the launch agency, as GCAT records it. 2026 is partial (to 2026 Oct 27). Hover or tap a year.
When the Soviet Union broke up, the launch rate fell with it. The Soviet Union launched 61 times in its final year, 1991; by the early 2000s the whole world managed little more than one a week, with a low of 54 in 2004. China's rise filled some of the gap in the 2010s. Then came reusable rockets. SpaceX, which had flown 728 orbital launches by 2026, made 170 of the 329 attempts in 2025 (52%), 122 of them carrying its own Starlink internet satellites. The world first beat the 1967 record in 2021, and 20% of all launches in seventy years have happened since 2020.
In 2025 the world launched 6.1 times as many rockets as in 2004, and more than half were SpaceX's.
Rockets reached space long before anything reached orbit. GCAT lists 4,334 launches before 1957, all suborbital. The first to pass 100 km was a German A-4, better known as the V-2, fired from Peenemünde on the Baltic coast in 1944 Jun (apogee 120 km). GCAT records 628 A-4 and V-2 launches from 1942 to 1952, 418 of them in 1944 alone, most as weapons against London, Antwerp and Paris; 75 captured rockets were later flown from New Mexico. The engineers and the rockets went east and west after the war, and their descendants launched Sputnik and Explorer 1. This page counts only the orbital launches that followed.
Of all 7,416 attempts, 507 (6.8%) failed to reach orbit. In the first ten years, to 1966, 22% failed. By the 1980s the rate was down to 4.2%, and in the 2020s it is 4.1%. 1958 was the worst year, with 20 failures. The newest launchers still fail more often: GCAT's smaller launching countries (Iran, Israel, the two Koreas, Brazil, Australia and others) have lost 43% of their 60 attempts. Turn on "Show" in the failures row of the map to see where they happened.
Failures per year, and the failure rate by decade
Bars: launches that failed to reach orbit, by year. Line: share of attempts that failed in each decade (right scale). Partial failures (some payloads in the wrong orbit) count as failures if GCAT codes them as one. Hover a bar.
Three places dominate the history. Plesetsk, a military site in the forests of northern Russia, has the most orbital launch attempts of anywhere: 1,686, 1,381 of them before the end of 1991, peaking at 70 in 1977. Baikonur in Kazakhstan, where Sputnik and Yuri Gagarin left from, has 1,558; Cape Canaveral, with the neighbouring Kennedy Space Center, 1,393; and Vandenberg in California, the American site for polar orbits, 938. Those four account for 75% of every launch in seventy years.
The busiest spaceports, ever and now
Orbital launch attempts, 1957 to 2026 (left) and in 2025 alone (right). Click a bar to see the site on the map.
The present looks very different. In 2025 Florida alone hosted 109 launches, 33% of the world's, and Vandenberg 66. China now spreads its launches over four inland and coastal centres, Jiuquan, Xichang, Taiyuan and Wenchang, plus ships at sea. Rocket Lab's pad on the Mahia Peninsula in New Zealand, which first launched in 2017, managed 17 launches in 2025, more than Baikonur's 6. Plesetsk, once the busiest site in the world, saw 9.
Every spaceport
All 46 sites with at least one orbital launch attempt. Click a row to see it on the map.
The Earth spins eastwards, and a rocket on the pad is already moving with it: 465 m/s at the equator, falling with the cosine of the latitude. At Kourou in French Guiana, five degrees north of the equator, the free boost is 463 m/s; at Cape Canaveral 409 m/s; at Baikonur 323 m/s; at Plesetsk only 212 m/s. Latitude matters even more for satellites bound for the geostationary ring over the equator, because a rocket's orbit is tilted to the equator by at least its launch site's latitude, and taking that tilt out later costs fuel. That is why Europe, and before it France, built its spaceport in South America: 70% of Kourou's launches went towards geostationary orbit, against 27% from the Cape, 18% from Baikonur and almost none from Plesetsk.
Latitude, the Earth's free push, and where rockets head
Spaceports with at least eight launches, by distance from the equator. Bar length: the eastward speed the Earth gives a rocket on the pad. Right: direction of successful launches to low and medium orbits, from the inclination of the first payload's orbit (under 75° eastward, 75° to 105° polar or sun-synchronous, over 105° westward). Hover a row.
Direction depends on geography as much as physics. Rockets cannot be allowed to drop their spent stages on towns, so each site has safe corridors over sea or empty land. From Cape Canaveral, with the Atlantic to the east, 97% of launches to low orbit head east. Vandenberg looks south over the Pacific, so 68% of its low orbit launches go into polar or sun-synchronous orbits. Israel launches westward over the Mediterranean, against the Earth's spin, so that spent stages fall into the sea rather than over its neighbours: all of Palmachim's launches with a known orbit are retrograde. Across all 5,266 successful launches to low and medium orbit with a known inclination, 65% went east, 33% polar and 1.4% westward, and just 0.8% reached an orbit tilted more than two degrees less than their site's latitude: the latitude sets a floor. (GCAT has no launch azimuths, so direction here comes from the orbit's inclination.)
GCAT's main catalogue lists 69,999 objects launched into Earth orbit or beyond: 27,852 payloads (satellites and spacecraft), 6,291 rocket stages, 7,922 components such as nose fairings and adapters, and 27,908 catalogued fragments of debris. 34,974 of them (50%) have already come back: 33,715 re-entered the atmosphere on their own as their orbits decayed, 1,076 landed or were recovered (crew capsules, film canisters, Space Shuttles), and only 183 were deliberately steered down. 33,397 are still up there, 10,644 of them Starlink satellites.
What came back to Earth, by year
Catalogued objects that re-entered, were deorbited or landed, by year and type. 2026 is partial. Hover a bar.
The busiest year so far was 2022, with 2,470 objects, swollen by debris: 1,379 of them (56%) were fragments of Kosmos 1408, an old Soviet satellite that Russia blew apart with a missile in an anti-satellite test in November 2021. In 2025, 1,912 came down, and 679 of those (36%) were Starlink satellites, which SpaceX retires by lowering their orbits so that they burn up within months. In 2019 the total was 314.
The fate of every catalogued object
All 69,999 objects in GCAT's main catalogue, by type and by what has happened to them. Hover a block.
Where do they fall? For most objects GCAT gives only a date. But for 2,136 of them (6% of everything that came down, 84% of them since 2020) it records a rough place of re-entry or landing. Turn on "Re-entry places" in Map layers to see them. The uncontrolled re-entries among them, 2,033 objects, are scattered across every ocean and continent between the latitudes their orbits reached: 76% fell over the sea, close to the 71% of the Earth that is ocean, and 497 over land. Nothing about them is aimed. Only 5.1% came down within 2,500 km of Point Nemo, a circle that covers 3.8% of the Earth's surface.
Point Nemo, the oceanic pole of inaccessibility, lies at 48.88°S 123.39°W in the South Pacific, about 2,688 km from the nearest land: Ducie Island in the Pitcairn group, Motu Nui off Easter Island, and Maher Island off Antarctica. Hrvoje Lukatela, a Croatian survey engineer, identified it in 1992 and named it after Jules Verne's Captain Nemo. Because nobody lives there and few ships pass, space agencies aim their biggest returning craft at the ocean around it: Wikipedia's article on the spacecraft cemetery says more than 260 spacecraft were disposed of there between 1971 to 2016, nearly 200 pieces of it Russian, and the International Space Station is planned to follow in 2031.
How the space stations came down
Every Salyut, Skylab, Mir and Tiangong station, with GCAT's status and its recorded place ("POR" is GCAT's shorthand for the Pacific Ocean region). Click a row to see it on the map.
GCAT shows the cemetery's history station by station. Salyut 1, 3, 4, 5 and 6 were all deorbited into the Pacific in the 1970s and early 1980s. Two did not go to plan: Salyut 7 re-entered on its own in 1991 and fell over Argentina, and America's Skylab, 34 tonnes, came down over Western Australia in 1979. Salyut 2 failed within weeks of launch in 1973 and also fell on its own. Mir was steered down on 23 March 2001. China's first station, Tiangong 1, slipped out of control and re-entered on its own in 2018, over the central South Pacific; Tiangong 2 was deorbited into the South Pacific in 2019. The regular visitors are cargo ships. Wikipedia names Russia's Progress, Japan's H-II Transfer Vehicle and Europe's Automated Transfer Vehicle; GCAT catalogues 183 Progress ships since 1978, 10 Japanese cargo ships since 2009 and all 5 ATVs (2008 to 2014).
Deliberate deorbits are rare. Of the 34,974 catalogued objects that have come back, 183 (0.5%) are recorded as deorbited on purpose. Even counting cargo ships, whose last plunge is logged in GCAT's separate event catalogues rather than here, almost everything that falls from orbit falls wherever its orbit happens to decay. Most of it burns up long before it reaches the sea.
Data. GCAT release 1.8.8 (data update 2026): the launch list (76,016 launches of every kind), sites, organisations, launch vehicles and the main object catalogue (69,999 objects). An orbital launch attempt is a launch GCAT codes as orbital or deep space; I leave out 11 rockets that exploded on the pad before launch and 6 aborts before lift-off, and I follow GCAT's judgement elsewhere (it counts some Starship test flights, for instance). A failure is any launch GCAT codes as failed, including partial failures. Who launched each rocket comes from the state of the first agency GCAT lists, which is why Rocket Lab's New Zealand company counts for New Zealand and Starsem's 31 Soyuz launches from Baikonur count for Europe.
Places. Spaceports are GCAT's launch sites, merged where a site was renamed or two sit side by side; coordinates are GCAT's, to about a kilometre for most sites. Direction comes from the inclination of the first payload's orbit in the catalogue, for successful launches to low and medium orbits only. The re-entry places are GCAT's own, written as rough coordinates (often whole degrees, some marked uncertain) for 2,136 objects; I do not infer a place for any other object. The 2,500 km circle around Point Nemo is my own device for drawing and counting: neither GCAT nor Wikipedia gives the cemetery a boundary.
What this is not. GCAT's main catalogue holds objects with standard catalogue numbers; it leaves out the newest and smallest objects that the US Space Force has not yet numbered, and a docked spacecraft's later fate (a cargo ship undocking and being deorbited, say) sits in GCAT's event catalogues, which I did not use. Debris counts are only the fragments big enough to be tracked. The figures for 2026 run to 2026 Oct 27.