Summary:
The European Space Agency (ESA) published its annual Space Environment Report on 8 September 2026, the agency’s yearly accounting of objects in orbit around Earth. The report treats near-Earth orbit as a limited resource, and its central warning is that some orbital regions could become increasingly unsafe and potentially unusable as debris accumulates.
At the end of 2025, ESA counted 44,964 objects in orbit, including 16,946 payloads — satellites and other spacecraft, working or dead — and 11,087 more detected without ever being tied to a launch or a break-up. About four payloads went up in 2025 for every payload that came down: 4,409 were launched, while 1,881 objects re-entered the atmosphere, 1,117 of them payloads.
The report’s projections show catastrophic collisions in low Earth orbit rising steeply after about 2075 if current launch and disposal behaviour continues. Even in a scenario with no further launches, debris keeps accumulating, because break-ups add fragments faster than they fall back to Earth. ESA’s space environment health index jumped from about 4 to about 50 in a single year, where 1 is the proposed sustainability threshold — the level of risk judged acceptable before the constellation boom — meaning risk 50 times higher than that level. The report says the figure should not be read as a deterministic forecast.
No spacecraft has yet captured and removed a large piece of debris that was not built to be caught. The first attempt, ADRAS-J2, is planned for a window running from April 2027 to March 2028, and ESA’s own removal mission is scheduled to launch in 2030.
Detailed Report
1. ESA’s Tenth Report Counts Orbital Traffic Through 2025
The Space Debris Office at ESA’s operations centre in Darmstadt, Germany, issued the report on 8 September 2026, and ESA published a public summary of it on 14 September. It is the tenth edition of a series that began in 2017. Its stated purpose is to give a transparent account of activity in orbit and to measure whether internationally agreed mitigation rules are holding. ESA says the orbits used most are filling with the remains of dead satellites and rockets.
The count recognises two kinds of object. A catalogued object is one whose orbit a surveillance system has tracked over a long period. An asserted object has not been tracked individually but is known to exist because it was built and launched — a rocket stage known to have separated from its launcher, for instance, even if no sensor has followed it since. Fragments too small or too short-lived for either list fall outside the count altogether, and ESA says both its debris and unidentified classes are almost certainly undercounts, because catalogued totals rise whenever surveillance improves.
The report divides near-Earth space into two regions. Low Earth orbit runs from the surface to an altitude of 2,000 kilometres. The geostationary region is a band roughly 400 kilometres thick centred about 35,800 kilometres above the equator, where satellites orbit in step with Earth’s rotation — the orbit used for communications and weather satellites. Most traffic sits in a few altitude bands, above all between 400 and 600 kilometres, where active, manoeuvrable satellites fly and coordination between operators matters most. Higher up, debris lingers for years or decades because the air is too thin to drag it down.
Only what has already flown enters the count; satellites still being prepared for launch are excluded. ESA cautions that its compliance measures are estimates that can shift between yearly releases.
2. Orbit Holds 44,964 Objects, and Most Risk Sits in Dead Hardware
Of the 44,964 objects in orbit at the end of 2025, 25,060 were in low Earth orbit, including 14,906 payloads. Payloads in all orbits numbered 16,946. Rocket bodies — the spent upper stages that carry satellites to orbit — numbered 2,080, and fragments of satellite and rocket break-ups came to more than 13,000. A further 11,087 objects, roughly a quarter of the total, have been detected but never tied to any launch or break-up. Debris has outnumbered working satellites since the space age began in 1957.
Known mass in orbit comes to 15,883 tonnes, 8,749.9 tonnes of it in low Earth orbit. That is a floor rather than a full measure, because the mass of most fragments is unknown and is entered as zero.
The report’s debris index scores each object by its potential to create fragments and trigger later collisions, and it puts the present risk almost entirely on hardware that no longer works. Inactive objects carry 96% of the index. Rocket bodies account for 65% of it and dead payloads for 30%. Large constellations carry 3% and working payloads 2%, on the assumption that 90% of active satellites are disposed of properly at the end of their missions. The risk concentrates around 850 kilometres in altitude, in orbits that pass near the poles at 70 to 80 degrees to the equator.
The report’s close-approach analysis shows debris from earlier break-ups feeding the same crowded shell. At 700 to 900 kilometres, the objects that pass close enough to prompt an operator’s collision check include fragments of the satellites Fengyun-1C, broken up in 2007, and Cosmos-2251 and Iridium 33, which broke up in 2009.
3. 2025 Set a Launch Record: 4,409 Payloads, Mostly Commercial
Launch traffic in 2025 was the highest the series has yet seen. More than 300 launches put 4,409 new payloads into orbit — 4,353 of them into low Earth orbit — along with 327 rocket bodies, a pace that ESA puts at about ten new payloads a day.
Rideshare flights, which carry many satellites on a single rocket, and large constellations explain the rate. More satellites now fly on each rocket, and single-payload missions are becoming rarer; by 2021, launching more than one payload per flight had become the most common way into orbit. The report counts 54 constellation or shell-level groups. A group qualifies if it has at least ten spacecraft, released over more than two launches and more than a year, and flown to operate as a system. In 2025 several new large constellations began systematic deployment in low Earth orbit.
The traffic is overwhelmingly commercial, and it is concentrated. ESA’s chart of 2025 launches into orbits between 200 and 1,750 kilometres shows commercial launches clustered tightly at about 500 kilometres. Civil and defence launches are small by comparison.
The bands are not holding still. ESA reports that constellations are changing altitude from year to year, and their active payloads are spreading across a wider range. Combined with the permanence of debris in low Earth orbit, that spread creates a significant risk of close approaches in the most used orbits.
4. Re-entries Rose in 2025, and Ground Risk Got Its First Measure
Re-entry traffic rose alongside launches. ESA counted 1,881 objects re-entering the atmosphere in 2025. About 1,200 were intact satellites and rocket stages, an average of more than three a day. Payloads made up 1,117 of the total, against the 4,409 launched.
ESA ties the rise to heavier launch traffic, high solar activity, and the practice, common among large constellations, of flying in low orbits from which satellites fall back quickly. Rocket stages are also being brought down on purpose more often: controlled re-entries of rocket bodies rose from about 10% to more than 65% over the decade, and for the second year running they outnumbered uncontrolled ones.
Re-entry carries a hazard of its own, and this edition measures it in aggregate for the first time. A fragment that reaches the ground carrying at least 15 joules of kinetic energy is counted as capable of severe injury. A one-in-10,000 chance of a casualty per re-entry is a widely used limit for a single object; the new measure adds the risk across all uncontrolled re-entries in a year. ESA’s chart ends 2025 near 0.045 expected casualties for rocket bodies and 0.037 for payloads, with constellation satellites under 0.01 once their safety design is counted. Controlled re-entries and human spaceflight are left out.
ESA says the risk to any individual remains extremely low, but expects the aggregate number of casualties to grow with the number of re-entries if safer design measures are not adopted. It points to more controlled re-entries and the wider adoption of design-for-demise — building satellites to burn up as completely as possible — as ways to reduce it.
5. Break-Ups Hold Steady, but Unidentified Objects Pile Up
Break-ups, not collisions, are still where most new fragments come from. Over the last two decades ESA counts an average of 9.8 non-deliberate break-ups a year, a rate it calls stable. Weighting the count by how long the pieces stay in orbit reduces the rate that still weighs on the environment to about 1.5 a year. Removing events ESA classes as systematic or unexplained lowers it again, to under one a year. The report adds that events with a large environmental impact are still occurring, in part because of designs already known to be problematic.
Much of the energy behind a break-up is stored power — fuel, batteries and flywheels — which can burst a spacecraft after its mission ends. The report’s mitigation objectives include passivation, which means draining that energy so a dead spacecraft cannot explode.
The year 2024 had been worse, with break-ups adding more than 3,000 catalogued fragments, mostly from propulsion-related failures. In contrast, 2025 brought only three break-ups. A 3,980-kilogram payload in medium Earth orbit broke apart on 17 January because of its propulsion system, producing 67 asserted pieces. A 2,020-kilogram rocket body fragmented on 1 June in what ESA classes as extended geostationary orbit, for a cause ESA lists as unknown, and a 530-kilogram payload in low Earth orbit followed on 18 December, again from propulsion failure. Most of the fragments from these events are asserted rather than catalogued: they were picked up by surveillance networks other than the United States Space Surveillance Network, and by commercial service providers. The American military system, which supplies much of the standard catalogue, registered only two objects across all three events, both from the June break-up.
The unidentified class is where the count is growing fastest. ESA added 3,509 unidentified objects in 2025, against 152 fragments traced to satellite break-ups and 43 to rocket break-ups. ESA describes a growing gap between objects detected and objects of known origin: sensors see more than they used to, but many detections cannot yet be linked to a source or a break-up event.
6. Collisions Could Sustain Debris Growth Even Without New Launches
The long-term concern is a chain reaction: a collision shatters an object into fragments, the fragments strike other objects, and each new collision adds more. Donald Kessler, a NASA scientist, and his co-author Burton Cour-Palais described the mechanism in a 1978 paper, and it carries his name: the Kessler syndrome. ESA calls this self-sustaining cycle, rather than launch numbers alone, the greatest long-term threat.
The cycle does not need new satellites to keep going. ESA says fragmentation events generate debris faster than it can naturally re-enter, so the population in low Earth orbit would keep growing even if launches stopped. Most debris large enough to cause a catastrophic collision — meaning above 10 centimetres — comes from the more than 500 break-ups of spaceflight history. Simulations indicate that collision fragments will begin to dominate at 800 to 1,400 kilometres within a few decades, with the runaway most likely to start near 900 and 1,400 kilometres, where debris is densest and air drag is too weak to clear it.
ESA’s projection chart shows the cycle at work. In the scenario that continues current launch and disposal behaviour, cumulative catastrophic collisions in low Earth orbit bend upward after about 2075 and reach roughly 1,750 by 2125. Last year’s projection ended near 250. The scenario with no further launches stays near the floor, low but still rising.
The same model projects how many objects larger than 10 centimetres, the size that can cause a catastrophic collision, would sit in low Earth orbit. The 2026 scenario, which continues current launch behaviour, climbs past 400,000 by the early 2120s, with the band of model results reaching about 450,000. The scenarios run from 2021 to 2025 stay below roughly 100,000, and those from 2005 and 2014 sit near 30,000. The projection horizon was cut from 200 years to 100, and the numbers are still higher than a year ago. ESA’s public summary describes the trend as “skyrocketing in a runaway effect.”
Both charts come from DELTA-4, ESA’s long-term debris model, which ran 100 simulations over 100 years from the population as it stood on 1 January 2026. The model repeats the launch traffic of 2017 to 2025, so the spread in its results reflects only the randomness of collisions and explosions, not uncertainty about future launch plans.
Disposal of satellites at the end of their missions is the first answer to this cycle. Of the payloads that reached the end of their missions in low Earth orbit since 2020, between 88% and 99% of those under 1,000 kilograms sit in orbits that bring them down within 25 years unaided. Only 54% of heavier payloads are set to remove themselves in that time. ESA’s own 2023 standard, binding on its projects, cut the limit to five years, and against that stricter threshold between 5% and 35% of payloads that ended life in a non-compliant orbit over the last decade complied. A common disposal target is 90%, and the report says that rate alone will not reduce the amount of debris; it must rise toward 100%. Studies by the Inter-Agency Space Debris Coordination Committee, which brings together 13 space agencies, put the requirement for constellations of small satellites in low orbit at 95% for a potentially stable environment.
7. The Health Index: From 4 to 50 in a Single Year
ESA condenses these projections into a single score, the Space Environment Health Index, which it introduced on 22 October 2025. A value of 1 is the proposed threshold for long-term sustainability. The benchmark is the world before the constellation boom: 2014 launch traffic combined with a 90% success rate in disposing of satellites within 25 years, a level of risk judged acceptable at the time.
At its introduction the index stood at about 4, a level ESA said was already well beyond the sustainability threshold. This year’s business-as-usual scenario puts it near 50, meaning risk 50 times higher than that first target, and ESA calls the change an order of magnitude in a single year.
The report stresses that the number is not a deterministic forecast. It shows how the environment could evolve on today’s knowledge and assumptions. The jump is driven by a much larger projected future population, built on constellation plans whose deployment, operating practices and end-of-life strategies are still maturing and on which verifiable detail is limited. The outlook could improve substantially or deteriorate further as behaviour becomes observable, and the scenarios with no further launches sit on the threshold line.
The report notes that a single constellation can substantially affect the long-term stability of the orbital environment. ESA says adherence to mitigation measures is rising on some indicators, but must become faster, more consistent and more reliable.
8. No Mission Has Yet Removed a Large Piece of Real Debris
ESA’s public summary says active debris removal is required to stop collisions from sustaining the growth, and that preventing new debris is no longer enough on its own. The flights and contracts so far fall into four groups. ADRAS-J inspected real debris without trying to catch it. ELSA-M will capture a satellite built to be serviced. ADRAS-J2, ClearSpace-1 and COSMIC are meant to capture targets never designed to be caught: a spent rocket stage, a working ESA satellite and two defunct British satellites that cannot cooperate. The United States awards cover disposal of the agencies’ own satellites. None has yet taken a piece of genuine debris out of orbit.
The closest approach to real debris to date belongs to ADRAS-J, built by the Japanese company Astroscale for the Japan Aerospace Exploration Agency and launched in February 2024. Its target was a spent H-IIA rocket stage left in orbit since the 2009 launch of the GOSAT greenhouse-gas satellite. In December 2024 ADRAS-J closed to within 15 metres of the stage, which Astroscale called the closest approach to space debris by a commercial company. Fly-arounds confirmed no major damage to the fitting on the stage where a follow-on mission would grab it. Astroscale said on 25 March 2026 that the spacecraft had begun deorbit operations after 293 days of mission life, having already lowered itself to an orbit from which it will re-enter within five years. It inspected the stage without capturing it.
The first capture attempt will be ADRAS-J2, the second phase of the programme, which returns to the same stage. Astroscale has chosen Isar Aerospace, a German launch company, to fly it on the Spectrum rocket from Andøya in Norway in Japan’s financial year 2027, which runs from April 2027 to March 2028. Spectrum reached orbit for the first time on its second flight on 5 September 2026, after the first, in March 2025, ended about 30 seconds after liftoff. Earlier demonstrations used targets the mission carried itself. RemoveDEBRIS, a mission led by the University of Surrey, netted a CubeSat it had released as artificial debris in 2018 and fired a harpoon at a target panel on a boom in February 2019, and Astroscale’s ELSA-d captured a client satellite launched with it, using magnets, in August 2021. ADRAS-J2 would be the first to capture and remove a large piece of real debris that was never designed for it.
ESA’s own attempt is ClearSpace-1, which the space company OHB System is developing with partners. Its target is PROBA-1, an ESA satellite still operating after more than 20 years in orbit. The satellite was never designed to be captured: it has no navigation aids or grappling points, and it tumbles. ClearSpace-1 must match that motion, seize the satellite and bring the pair down to re-enter within three years. A 1 July briefing by the mission’s developers at ESA’s Clean Space Days set launch for the second quarter of 2030, with re-entry in 2034.
Two further efforts are earlier in development. ELSA-M, a servicer from Astroscale UK, will first go after a retired satellite from the Eutelsat OneWeb constellation, one built to be serviced. A separate British project, COSMIC, is designed to capture two defunct United Kingdom-registered satellites that cannot cooperate. Neither has flown.
In the United States, the Space Development Agency and the Defense Innovation Unit, two Defense Department bodies, awarded a combined US$8.4 million in August to Firefly Aerospace, D-Orbit and Katalyst Space to design deorbit-as-a-service vehicles. These are spacecraft that would rendezvous with a satellite at the end of its life, capture it and pull it down to burn up. Preliminary design reviews are due by the end of 2026. The Space Development Agency signed Starfish Space in January 2026 to a contract valued at up to US$52.5 million, to dispose of as many as seven of the agency’s own spacecraft. Starfish plans to launch its Otter vehicle for that work in 2027. Both sets of awards cover retiring the agencies’ own satellites, not clearing debris already in orbit.
Conclusion
ESA’s report counts 44,964 objects in orbit at the end of 2025 and projects collisions in low Earth orbit climbing after about 2075 if current launch and disposal behaviour continues. Its health index rose from about 4 to about 50 in a year, a figure the report says should not be read as a deterministic forecast.
No spacecraft has yet captured and removed a large piece of debris that was not built to be caught. As of 8 October 2026, ADRAS-J2 is planned to launch between April 2027 and March 2028 and ClearSpace-1 in 2030. The next dated event is a design review, due by the end of 2026, of United States contracts to retire the agencies’ own satellites. The first attempt to capture debris already in orbit is ADRAS-J2, whose window opens in April 2027.