3,456 Recalls, Zero Trampoline Courts: What the Safety Record Names
On 5 November 2026 the Consumer Product Safety Commission published a recall of 10,318 Arlopu slide playsets, because a child’s head or neck could be caught where the slide meets the platform. That record is one of 3,456 the agency issued between 1 January 2016 and 21 November 2026, and every one of them is public, dated and searchable by product description.
Search the same 3,456 records for “trampoline park”, “trampoline court”, “indoor playground”, “ball pit”, “foam pit” or “obstacle course” and the count is zero for every phrase. Search for “auto belay” and five records appear, from three manufacturers, between March 2020 and February 2025.
The first reaction is usually that trampoline parks must be safe and climbing walls must not be. That reading does not survive contact with the injury data, which records more than a hundred thousand trampoline-related emergency visits in a single year in the United States. The more useful reading is structural. A recall is an instrument aimed at a product, and a park is not a product. It is a building, an installation and a set of procedures, and none of those three is something the Commission can order back to a factory.
What follows is the record itself, sorted into three classes, with the arithmetic shown. It is a map of what a recall can and cannot tell a buyer, and it ends with three documents worth asking a supplier for instead of a recall search.
1. What was counted, and what the count covers
The dataset is the Commission’s public recall service, queried for the window from 1 January 2016 to 21 November 2026. It returned 3,456 records. Each record carries a recall number, a date, a title, a narrative description, a product name, an estimated unit count and a link to the agency’s own notice. The query is reproducible by anyone with the endpoint and the two date parameters, and the count is not an estimate.
Sorting those 3,456 records into play equipment takes a second pass, because a keyword search on its own produces false positives. “Slide” also catches slide-in ranges and neckerchief slides. “Play set” also catches toy playsets and magnetic fidget sliders. The list below is a curated subset of 28 records, each read and kept only where a child climbs, slides, swings or bounces on the product. The split into three classes is my own reading of the product descriptions rather than an agency taxonomy, and every boundary is stated in the caveats further down.
Two numbers anchor the whole exercise. Twenty-eight records out of 3,456 is 0.81%. The park-relevant hardware alone, meaning the nine institutional play structures plus the five belay devices, is fourteen records, or 0.41%.
2. The 28 records that touch play equipment
The institutional class is the one a venue owner recognises. These are products sold to parks, schools, municipalities and family entertainment centres rather than to households.
| Date | Company | Product | Hazard named | Units |
|---|---|---|---|---|
| 2016-12-07 | Playworld | Stainless steel playground slides | Amputation | 1,300 |
| 2018-05-16 | Play and Park Structures | Playground slides | Entrapment | 150 |
| 2018-10-04 | BCI Burke | Metal playground climbers | Strangulation | 200 |
| 2019-10-18 | Soft Play | Spider Climb climbers | Fall | 200 |
| 2019-12-19 | BCI Burke | Merge playground climbers | Entrapment | 440 |
| 2023-07-27 | Soft Play | Sky Wheels | Fall | 3,500 |
| 2025-05-29 | Playworld Systems | Unity Teeter Tunnels, Lexan tops | Laceration, crush | 329 |
| 2026-03-12 | LFTE USA | Playground swing set seats | Fall | 7,200 |
| 2026-10-08 | Pacearth Direct | Swing seats | Fall | 110,550 |
The belay class is smaller and, for anyone operating a climbing wall or a high-ropes element, more pointed. Every one of the five records names a fall hazard on a device whose only job is to arrest a fall.
| Date | Company | Product | Hazard named | Units |
|---|---|---|---|---|
| 2020-03-19 | C3 Manufacturing | Perfect Descent belay devices | Fall | 1,100 |
| 2021-03-24 | Beal SAS | Birdie belay devices | Injury | 1,400 |
| 2024-06-20 | Head Rush Technologies | TRUBLUE iQ auto belays | Fall | 2,200 |
| 2024-08-22 | Head Rush Technologies | TRUBLUE Speed auto belays | Fall | 220 |
| 2025-02-27 | Head Rush Technologies | TRUBLUE iQ and iQ+ auto belays | Fall | 410 |
The consumer class carries the volume. These products reach households through retail and e-commerce channels, and the unit counts are an order of magnitude larger than anything in the institutional list.
| Date | Company | Product | Hazard named | Units |
|---|---|---|---|---|
| 2016-04-07 | Rainbow Play Systems | Plastic yellow trapeze rings | Reannounced, low response | 121,000 pairs |
| 2019-02-26 | JumpSport | Mini trampolines | Injury | 11,300 |
| 2019-08-01 | Super Jumper | Trampolines | Fall, injury | 23,000 |
| 2020-05-14 | Flybar | Swurfer baby and toddler swings | Fall | 3,000 |
| 2021-05-13 | Leisure Time Products | Brutus swing sets | Injury | 9,000 |
| 2021-11-04 | Backyard Play Systems | Playsets with wooden roof | Entrapment | 5,100 |
| 2022-07-28 | Epoch Everlasting Play | Kidoozie play tents and playhouses | Flammability, burn | 251,600 |
| 2023-10-12 | The FORT | Children’s play tents | Choking, laceration | 13,250 |
| 2026-02-26 | TheKiddoSpace | Children’s sensory swing | Strangulation | 140 |
| 2026-04-16 | SEGMART | Mini round toddler trampolines | Strangulation | 1,200 |
| 2026-04-16 | Fun and Function | Metal multi-point swing frames | Fall | 1,300 |
| 2026-06-04 | Tiny Land | Children’s play tents | Skin and eye irritation | 7,500 |
| 2026-10-29 | Harkla | Children’s sensory swings | Fall | 25,541 |
| 2026-11-05 | Arlopu | Slide playsets | Head and neck entrapment | 10,318 |
Unit counts are the leading figure printed in each agency notice. Several records carry a second quantity described as additional units, and the Rainbow rings were sold in pairs, so the column is a scale indicator rather than an exact total. Nothing in it should be added to anything else.
3. What the counts turn into
The three classes differ in count and differ far more in batch size. That difference is the first real finding, because it tells you what kind of event a recall actually is.
| Class | Records | Share of 3,456 | Units at stake | Median record | Dominant hazard |
|---|---|---|---|---|---|
| Institutional play structures | 9 | 0.26% | 123,869 | 440 units | Fall, entrapment |
| Belay and descent devices | 5 | 0.14% | 5,330 | 1,100 units | Fall |
| Consumer play products | 14 | 0.41% | 483,249 | 10,318 units | Fall, entrapment, strangulation |
| All three classes | 28 | 0.81% | 612,448 | 2,600 units | Fall |
The median institutional recall covered 440 units. The median consumer recall covered 10,318, which is twenty-three times as many. Average batch size tells the same story in one line: 13,763 units per institutional record, 1,066 per belay record, 34,518 per consumer record.
That gap is not a coincidence of this sample. A commercial play structure is specified per site and installed by a contractor, so a defect reaches hundreds of sites at most. A consumer playset is moulded, boxed and shipped to a warehouse, so a defect reaches a national retail channel. The recall instrument works well on the second case and awkwardly on the first.
The hazard column is where the two worlds converge. Across all 28 records, fall is named twelve times, entrapment four times and strangulation three times; the other nine name amputation, laceration, flammability, choking, irritation or injury without further detail. Those first four categories are also the four failure modes a park operator writes into a daily inspection sheet. The recall record is not telling a venue owner anything new about physics. It is telling them which components carry the risk.
The arithmetic in one line: 28 divided by 3,456 is 0.0081. Fourteen park-relevant records divided by 3,456 is 0.0041. The record is thin, and the thinness is the point.
4. The attachment is what gets recalled
Read the 28 product names in sequence and a pattern appears that no single recall notice states. Not one record names a whole venue. Every record names a component: a slide, a climber, a teeter tunnel, a swing seat, a swing frame, a play tent, a belay device. The unit of recall is the part you bolt on or hang from, not the floor you build.
The belay records make this concrete. Five recalls between March 2020 and February 2025, from three manufacturers, all naming a fall hazard on a device that exists to arrest falls. An auto belay sits at the top of a climbing wall and pays out a lanyard as a climber ascends. It is the single point where a climber’s weight is held by a machine rather than by a person. In a family entertainment centre it usually sits on one wall at the back of the hall, well outside the sightline of the front desk, and it is often the least examined piece of hardware in the building.
A device recall and a procedural failure are separate paths to the same outcome, which is why both belong in a pre-opening checklist. The Victorian prosecution that followed the 2022 fall at a Bounce venue in Grovedale turned on a carabiner left parked on a floor mat, not on a defective device, and the measures the court described were procedural: a separate harness fitter, a documented activity plan, no unaccompanied children under twelve. We worked through that case and its 1:15 ratio in the supervision ratio analysis. The recall record adds the other half: the machine at the top of the wall has its own history, and it is public.
The institutional structures point the same way. Three of the nine are climbers, two are slides, one is a teeter tunnel and three are swing or trapeze hardware. Every one of those is an interface between a child’s body and a hard edge, a moving part or a gap. That is exactly the list a design review should walk through before a layout is signed, and it is why the standards governing these products are written around geometry rather than around brand.
5. Why the trampoline court is not in the record
The absence needs an explanation, and the explanation is jurisdictional rather than practical. The Commission’s authority runs over consumer products. A trampoline court welded, installed and operated inside a commercial venue is not a consumer product in that sense. It is a fixture of a business, and the compliance route for it runs through voluntary consensus standards, state and local amusement-device codes and the insurer’s inspection, not through a federal recall docket.
The second reason is mechanical. A recall assumes a manufacturer who can reach a population of identical units and remediate them. A court is built once, on one slab, to one drawing. When something is wrong with it, the fix happens on site, by the installer, under a contract, and it never generates a public notice. The absence of a recall is therefore evidence of how the equipment is sold, not evidence of how it performs.
Two more probes make the same point. “Soft contained play”, the phrase that describes the padded multi-level structures common in soft play centres, returns zero records across the whole window, even though a dedicated consensus standard exists for exactly that product class. “Ball pit” returns zero. “Foam pit” returns zero. These are not claims that nothing has ever gone wrong in a ball pit. They are claims that the federal recall docket is not where those events are recorded.
Where the events are recorded is injury surveillance and enforcement. The 2024 emergency-department analysis we worked through earlier found 101,474 trampoline-related visits in a single year, with oral, facial and head injuries concentrated in the youngest age band; the odds are set out in the NEISS age analysis. Staff injuries sit in a different dataset again, at 6.4 recordable cases per 100 workers for the amusement and recreation code, as set out in the staff injury comparison. Recalls are a third instrument, and the three do not measure the same thing.
One boundary belongs here rather than in the caveats. Everything above describes the United States federal system. The European Union’s Safety Gate, the United Kingdom’s product safety regime and Australia’s recall system each publish their own notices on their own schedules, and none of them is included in these figures. A buyer in Manchester or Melbourne should read this as a method to repeat, not as a result to import.
6. What the 2026 share is made of
Eight of the 28 records fall inside 2026, up to 21 November. The previous ten full years produced twenty, or two per year. Eight records in roughly ten and a half months annualises to about nine, which is four and a half times the earlier rate. On a sample this small, that ratio should be read as a flag rather than a trend.
What the eight are matters more than how many. Six are consumer-channel products from importers and e-commerce brands: two play tent recalls, three sensory swing recalls and one toddler trampoline recall. Two are swing-seat recalls with much larger unit counts, 7,200 and 110,550. None of the eight names a trampoline court, an indoor playground or a soft contained play structure.
Two explanations fit the numbers equally well, and this data cannot separate them. The first is that the consumer channel has genuinely expanded, with more small importers placing more moulded play products into a market that watches them more closely than it did in 2016. The second is that reporting and filing practice has changed, so defects once handled by a retailer’s withdrawal now become formal notices. Both would produce the pattern above. A reader who wants to test which one is operating should watch whether the 2027 records keep arriving from the same channel.
7. What the data does and does not show
This section exists because the numbers above are easy to misuse, and the misuse would be flattering to nobody.
It does not show injury rates. A recall is an administrative event triggered by a defect pattern, a manufacturer’s decision and a regulator’s agreement. Products with no recall can still injure people, and a product with a recall may have caused no injury at all before it was pulled. Nothing here should be read as a ranking of safety by manufacturer or by category.
The 28-record subset is a keyword search plus a manual read. The window contains 3,456 records, and a phrase search returns both the product and its neighbours. Products whose primary function is not play, climbing, sliding or bouncing were excluded by hand, and a different reviewer would place the boundary slightly differently. The three-class split is my own construction and does not correspond to any official category.
Unit counts are approximate and not additive. Notices print “about” a number, sometimes with an additional quantity in a second sentence, and the Rainbow rings were sold in pairs. The 612,448 total is a scale indicator built from leading figures only.
There is no denominator. The dataset says how many recalls happened, not how many parks, courts or play structures exist, so no rate per venue can be computed from it. Any statement of the form “one recall per X parks” would need an installed-base figure this data does not contain.
The window is partial at both ends. It starts on 1 January 2016 and stops on 21 November 2026, so 2026 covers ten and a half months rather than a year, and anything published after the query date is missing. Records are counted by the date the agency published them, not by the date of the incident that prompted them.
The jurisdiction is one country. These are United States federal records. Canada, the European Union, the United Kingdom and Australia maintain separate systems with separate thresholds for what gets published, and a product recalled in one market is frequently sold without notice in another.
8. What to ask a supplier instead
A recall search answers a question that is almost never the real one. The question buyers need answered is which component in the building has a history, and what evidence proves that the specific unit on their slab was built and installed to the standard that history points at. Three documents do that work.
Ask one A component list with a standard and an edition against each line, not a brand-level certificate. A trampoline court, a netting system, a padding set and a belay device are governed by different documents, and a single certificate that claims to cover all four is telling you it covers none of them precisely. The scope paragraph is the part to read, not the logo.
Ask two The in-place test record for anything whose performance changes after installation. Netting is the clearest example: a panel that passes a factory test can still be tied off at the wrong spacing on site, and the in-place test is the only thing that catches it. The parameters and the arithmetic are in the netting specification breakdown.
Ask three A replacement schedule for the wear items that appear in the recall classes above: swing seats, webbing and straps, net panels, pad covers and the lanyard on any belay device. A schedule with intervals and part numbers is a procurement document. A schedule that says “as required” is not.
| If your venue has | Verify this first | Because the record shows |
|---|---|---|
| A climbing wall or high-ropes element | Belay device model, recall status, inspection interval and the log | Five fall-hazard recalls on belay devices since 2020 |
| A trampoline court | Frame and leg load design, net attachment detail, pad coverage | No federal court recall exists to rely on |
| Soft contained play | Entrapment geometry at slide entries, fire treatment of covers | Entrapment and flammability dominate the play records |
| Swings or suspended seats | Seat and strap specification with a replacement interval | Two 2026 swing-seat recalls, 117,750 units |
The pattern across all 28 records is consistent enough to state as a rule. Recalls attach to parts, and parts are what a buyer can specify. A venue that specifies at component level, keeps the in-place test records and holds a dated replacement schedule is doing the thing the recall system cannot do for it, which is to close the loop between what was certified and what is bolted to the floor.
If you are at the specification stage, the practical next step is a component schedule for your own layout rather than a folder of certificates. The design and delivery process sets out how that schedule is produced, and the safety standards page explains how certificates are scoped per component. For a review against your own building, send the plan and the clear height and we will mark up what the shell allows.
FAQ
Has a trampoline park ever been recalled in the United States?
None appears in the federal recall docket. Across 3,456 records published between 1 January 2016 and 21 November 2026, the phrases “trampoline park”, “trampoline court”, “commercial trampoline” and “indoor playground” return zero matches. The three trampoline records in that window are consumer products: a mini trampoline, a backyard trampoline and a toddler trampoline. A commercial court is installed as a fixture of a business, which places it outside the consumer-product route that produces recalls.
Does a CPSC recall cover commercial play equipment?
It can, when the product is sold as a product rather than installed as a fixture. The nine institutional records in this window are all of that kind: playground slides, climbers, a teeter tunnel and swing seats sold to parks, schools and contractors. What the docket does not contain is a whole venue, because a venue is not a unit that can be returned, refunded or fitted with a repair kit.
How do I check whether a component in my park has been recalled?
Search the agency’s public recall database by product description and by manufacturer name, then check the same model against the equivalent systems in the markets you operate in, since each jurisdiction publishes separately. Record the model and serial number of every belay device, and keep the search date with the record. A recall search is a point-in-time check, so the date matters as much as the result.
Which components appear most often in recall records?
In this window the recurring components are swing seats and swing frames, climbers, slides, play tents and belay devices. Fall is the most frequently named hazard, appearing in twelve of the 28 records, followed by entrapment and strangulation. Recalls attach to discrete parts rather than to complete structures, which is also why a component-level specification is more useful than a brand-level certificate.
Is ASTM F2970 certification required by law?
No. ASTM F2970 is a voluntary consensus standard for trampoline courts, and ASTM F1918 plays the same role for soft contained play equipment. What is legally required depends on the jurisdiction: state or local amusement-device codes, building and fire codes, and in some markets a specific national standard. A supplier certificate is evidence about a product; it is not a permit to open, and the two are checked by different people.
Sources
- U.S. Consumer Product Safety Commission, recall service, records published 1 January 2016 to 21 November 2026, queried 20 November 2026. saferproducts.gov/RestWebServices/Recall
- CPSC, “Playworld Recalls Stainless Steel Playground Slides Due to Amputation Hazard”, recall 17718, 7 December 2016. cpsc.gov
- CPSC, “Playworld Systems Recalls Unity Teeter Tunnels with Lexan Tops Due to Risk of Serious Injury and Crush Hazard”, recall 25303, 29 May 2025. cpsc.gov
- CPSC, “BCI Burke Recalls Merge Playground Climbers Due to Entrapment Hazard”, recall 20044, 19 December 2019. cpsc.gov
- CPSC, “Head Rush Technologies Recalls TRUBLUE iQ and TRUBLUE iQ+ Auto Belays Due to Fall Hazard”, recall 25165, 27 February 2025. cpsc.gov
- CPSC, “Head Rush Technologies Recalls TRUBLUE iQ Auto Belay Devices Due to Fall Hazard”, recall 24277, 20 June 2024. cpsc.gov
- CPSC, “Pacearth Direct Recalls PACEARTH Swing Seats Due to Risk of Serious Injury from Fall Hazard”, recall 27019, 8 October 2026. cpsc.gov
- ASTM International, F2970 Standard Practice for Design, Manufacture, Installation, Operation, Maintenance, Inspection and Major Modification of Trampoline Courts. astm.org
- ASTM International, F1918 Standard Safety Performance Specification for Soft Contained Play Equipment. astm.org
