Daniel sent us this one — and it's the question you're not supposed to ask out loud in an airport lounge. We've done episodes on the bureaucracy of safety, on automation dependency, on why the system mostly works. Daniel wants the flip side. Which airline has the single worst safety record on the planet right now? And which airspace has the most unreliable infrastructure — the place where the system itself is the hazard?
The timing is actually pointed. The EU Air Safety List got its June update, the blacklist — a hundred twenty-nine airlines banned from European airspace as of now. Meanwhile the Aviation Safety Network's data through midyear shows twenty twenty-five was the deadliest year for commercial aviation since twenty eighteen. So the numbers are fresh, and they are not subtle.
Let's start with a hard question. How do you even measure which airline is the most dangerous? Because the instinct is to Google "most plane crashes" and call it a day, but that's not how this works.
The metric that actually matters is fatal accidents per million flights — or the ratio of hull-loss accidents to departures. If you just count total crashes, the biggest airlines with the most planes win by default. American Airlines has had more accidents in absolute terms than some tiny Nepalese carrier, but American also operates something like two million flights a year. The denominator eats the risk. What you want is the rate.
A small airline that crashes once a decade can look worse on paper than a giant that crashes twice a year. The math is brutal.
And we're also scoping this deliberately. We're excluding cargo-only operators — different regulatory regime, different risk profile. We're looking at scheduled passenger airlines. And we're distinguishing between two things that get collapsed together all the time: the airline itself, and the airspace it flies through. They are not the same answer, and the interaction between them is where the real danger lives.
Two-part structure. Part one — the airline. Part two — the airspace. And the answer, it turns out, depends on which question you're asking.
Let's start with the airline. According to the Aviation Safety Network database, updated through mid twenty twenty-six, the scheduled passenger carrier with the highest fatal accident rate per million flights is Yeti Airlines in Nepal. Three point one two fatal accidents per million flights. For context, the global average across all scheduled carriers is below zero point five.
Three point one two. That's not a rounding error — that's six times the global baseline.
It's driven largely by one catastrophic event that everyone remembers. January fifteenth, twenty twenty-three. Yeti Airlines Flight six ninety-one, an ATR seventy-two five hundred, crashed on approach to Pokhara International Airport. Seventy-two people killed.
That crash had a detail that, when I first read the preliminary report, I genuinely thought I'd misread it. The pilots feathered both propellers.
Both of them. And for listeners who don't live inside a cockpit, let me explain what that means and why it's so damning. On a turboprop like the ATR seventy-two, feathering the propeller means rotating the blades so they're edge-on to the airflow — it eliminates drag if an engine fails. It's what you do when an engine quits. You feather the dead engine's prop. You do not feather the working engine. And you absolutely, under no circumstances, feather both.
What they did was the aerodynamic equivalent of turning off both engines at five hundred feet while trying to climb.
The cockpit voice recorder showed the sequence. The aircraft was on final approach, the crew initiated a go-around — they pushed the power levers forward — but then the first officer inadvertently moved the condition levers to the feathered position instead of the flight idle position. Both propellers feathered. The aircraft lost all thrust. It stalled, rolled left, and went into a ravine. No checklist was called, no challenge-response verification. The entire go-around procedure was executed from memory, incorrectly.
That's not a mechanical failure. That's a training failure so fundamental it's almost hard to process. The difference between the condition lever and the power lever is day-one ground school on a turboprop.
This wasn't Yeti's first hull loss. Since two thousand, the airline has had four hull-loss accidents with eighty-nine total fatalities — across a fleet of roughly fifteen aircraft. That's a loss rate that would ground any carrier in a jurisdiction with functioning oversight. Yeti was already on the EU blacklist before the Pokhara crash. It's still flying domestic routes in Nepal today.
The question becomes: why does an airline with that record keep operating? And the answer is the regulatory environment. Nepal's Civil Aviation Authority has been flagged by ICAO for Significant Safety Concerns — that's the formal designation, SSC — since twenty thirteen. That's thirteen years of documented, unresolved safety deficiencies. And they continue to certify airlines.
The mechanism here is regulatory capture. The CAA in Nepal is not an independent safety body in the way the FAA or EASA is structured. It's embedded in the Ministry of Culture, Tourism, and Civil Aviation. It both promotes Nepalese aviation and regulates it — which is a structural conflict of interest. When tourism revenue depends on airlines operating, the regulator has an incentive to look the other way.
The terrain doesn't help. Pokhara, Lukla, Kathmandu — these are manual-flying approaches through mountain valleys with minimal radar coverage below ten thousand feet. You can't just engage the autopilot and let the ILS glide you down. There is no ILS at most of these airports. You're hand-flying a turboprop through terrain that gives you zero margin for error, and your training didn't teach you which lever does what.
That's the airline-level story. But I want to complicate the picture, because the ranking shifts depending on how you slice the data. Yeti tops the per-flight fatality rate among scheduled carriers with more than ten years of operations. But if you look at absolute fatality counts, larger airlines with systemic maintenance issues — Lion Air, Air India Express — have higher body counts. Lion Air had the six hundred ten crash in twenty eighteen, the Boeing seven thirty-seven MAX that went into the Java Sea, one hundred eighty-nine fatalities. Before that, a two thousand four crash in Solo, twenty-five dead. Multiple other incidents. But Lion Air operates a huge fleet — hundreds of aircraft — so its rate is lower than Yeti's even though its safety culture has been documented as problematic for years.
Then there are airlines most people have never heard of that are statistically worse than Yeti if you look at a different metric. Kam Air in Afghanistan. Blue Wing Airlines in Suriname. These are small carriers where a single crash produces an enormous per-departure fatality rate because the denominator is tiny.
Kam Air Flight nine oh four, February third two thousand five. A Boeing seven thirty-seven two hundred — already an aging aircraft — lost in a snowstorm near Kabul. One hundred four fatalities. The airline never recovered from the reputational damage, and it's been on the EU blacklist essentially since the blacklist existed. Afghanistan has no functioning civil aviation authority in any meaningful sense. ICAO can't even conduct a proper audit there — the USOAP audit program has had no access to Afghanistan's CAA for years.
Let me ask the obvious question. If the EU blacklist bans a hundred twenty-nine airlines, and Yeti is on it, and Kam Air is on it, and every airline from Nepal, Afghanistan, and the DRC is on it — is the blacklist actually a reliable proxy for danger? Or is it a political document dressed up as a safety tool?
It's both, and that's what makes it tricky. The blacklist is explicitly a political instrument — it's the European Commission deciding which airlines it will not allow into European airspace. The criteria include safety data, but also whether the airline's home regulator provides adequate oversight. An airline can have zero accidents and still be banned because its national CAA is a paper tiger. Conversely, an airline can be off the blacklist — Lion Air was removed in twenty sixteen — and still have serious safety culture problems.
The blacklist is a starting point, not a finish line. If an airline is banned from Europe, it's almost certainly unsafe or operating under a failed regulator. But if it's not banned, that doesn't mean it's safe. It just means Europe hasn't gotten around to banning it, or the politics don't align.
And that's the segue to part two, because the airline is only half the story. The airspace it flies through can make a safe airline dangerous — or a dangerous airline lethal.
Shift to airspace. The most dangerous airspace in the world, by accident rate per million movements, is the Central African region — the ICAO AFI region, specifically the Democratic Republic of Congo, South Sudan, and Somalia. What's the number?
Seven point four accidents per million departures, according to twenty twenty-four ICAO data. The global average is zero point nine. So African airspace is roughly eight times more dangerous than the global baseline. And within that, the DRC is the worst national performer: eleven point two accidents per million departures. That's not a typo. Eleven point two.
Eleven point two per million versus zero point nine globally. That's not a gap, that's a different reality.
The causes are not mysterious. They're just expensive to fix and politically inconvenient. Let me walk through the four structural failures. First: radar coverage. Over sixty percent of African airspace has no primary radar coverage below twenty-four thousand feet. When an aircraft descends below flight level two four zero, air traffic control loses it. They rely on procedural separation — time-based estimates of where aircraft should be — rather than radar-based tracking. And procedural separation works fine in theory but breaks down instantly with weather deviations, reroutes, or a pilot who's five minutes late on a position report.
You've got aircraft descending through clouds, into mountainous terrain, and the controller is essentially guessing where they are based on when they said they'd be somewhere ten minutes ago.
Second: ATC staffing and language. Many towers in the region have single controllers working twelve-hour shifts. The ICAO standard for English proficiency is Level Four — operational. In multiple African states, controllers are operating at Level Three or below, which means they can communicate in routine situations but break down under stress or non-standard phraseology. When an emergency happens, the controller and the pilot may not understand each other.
The third factor?
Forty percent of runways in sub-Saharan Africa lack precision approach aids. No PAPI lights. In some cases, no approach lighting at all. Pilots are flying visual approaches at night, in weather, to runways that may be unpaved. The Hewa Bora Airways crash in Goma in two thousand eight — a Boeing seven twenty-seven crashed on takeoff after an engine failure. Forty-four fatalities. The runway was unpaved, no approach lights, and the aircraft had no business operating in those conditions.
Hewa Bora is actually the case study for the fourth factor, which is regulatory collapse. Three fatal accidents between two thousand eight and twenty eleven. All involving approach or landing in poor weather with minimal ATC support. And the DRC's civil aviation authority had essentially no enforcement capacity. The airline kept flying until it couldn't.
That brings us to the interaction effect — what I think is the real insight here. The worst airline flying in the worst airspace isn't additive risk. It's multiplicative. A well-maintained European carrier flying into Juba, South Sudan is still riskier than flying a local carrier within the United States, because the airspace itself degrades safety margins. But a poorly maintained local airline flying in that same degraded airspace — that's where the fatality rate spikes off the chart.
The airline is the symptom, the airspace is the disease. You can have a safe airline operating in dangerous airspace and the risk is elevated but manageable. You can have a dangerous airline operating in safe airspace and the risk is contained — the system catches errors before they become crashes. But when both fail simultaneously, you get Hewa Bora. You get the DRC's accident rate at eleven point two.
Let me give you a recent example that illustrates the airspace failure in isolation. In twenty twenty-four, at Juba International Airport in South Sudan, a single air traffic controller managed all arrivals and departures for fourteen hours without relief. Two near-misses occurred in the same shift. This was not a conflict zone — South Sudan was not at war at that moment. This was just normal operations in a country where the CAA doesn't have the budget to staff towers properly.
A single controller. That's not a safety system, that's a hope-based approach to aviation.
It connects to something we've talked about before — the automation dependency issue. When pilots are used to flying in radar-rich environments with full ATC support, and then they're dropped into procedural airspace with degraded infrastructure, the skills gap becomes lethal. The Yeti Pokhara crash was a manual flying error. The DRC accidents are overwhelmingly approach and landing events where the crew is navigating without aids and without controller support.
What's the worst combination? If you had to name the single most dangerous seat on a commercial aircraft today, what is it?
A domestic flight on a small turboprop operated by a locally-owned airline in the DRC or Nepal, flying an approach to an airport without ILS, in marginal weather, with a single controller working a double shift and no radar coverage below ten thousand feet. That's not a theoretical worst-case. That's a Tuesday in Goma.
Yet people fly these routes every day because they have to. There's no alternative. You can't take a train across the Nepalese Himalayas.
Which is why the practical question matters. If you're a traveler who has to fly into high-risk airspace — and I mean has to, not vacation-choosing — what do you do? The EU blacklist is a starting point. Cross-reference it with the Aviation Safety Network's accident database, which is publicly available and searchable by airline. Look at ICAO's Universal Safety Oversight Audit results — USOAP scores are published and they tell you whether a country's regulator is functional or fictional.
The simplest heuristic: fleet age. An airline operating aircraft less than ten years old in a high-risk region is almost certainly better capitalized, better maintained, and subject to more external oversight than one flying thirty-year-old hand-me-downs. It's not a guarantee, but it's a strong signal.
IOSA certification — that's the IATA Operational Safety Audit — is another filter. Airlines that pass IOSA have a safety management system that's been externally verified. It's not perfect, but airlines without it are essentially self-certifying. And in a region with a captured regulator, self-certification means no certification.
The toolkit is: EU blacklist first — if the airline is banned, don't get on it. Then ASN accident history. Then USOAP scores for the country. Then fleet age and IOSA status. None of these alone is sufficient, but together they give you a picture.
Here's the counterintuitive part. The safest choice in a high-risk airspace is often not the local carrier. It's a foreign airline with a strong safety culture flying into that airspace. Ethiopian Airlines, for example, has an excellent safety record and operates throughout Africa with modern aircraft and proper training. Kenya Airways similarly. You pay more, but the safety margin is real.
That's the practical takeaway. But I want to zoom out for a second, because there's a structural question hanging over all of this. Aviation in Africa and South Asia is projected to grow at five and a half percent annually through twenty thirty. That's more traffic, more aircraft, more approaches into airports that already lack ILS and radar. Is the infrastructure keeping pace?
Right now, no. The investment gap is enormous. The African Development Bank estimated a few years ago that the continent needs something like sixty billion dollars in aviation infrastructure investment by twenty thirty just to maintain current safety levels — not improve them, just prevent backsliding. And that's not happening. What is happening is piecemeal upgrades funded by foreign aid or airport privatization deals that prioritize passenger terminals over radar installations, because a shiny new terminal is politically visible and a radar array isn't.
The political economy of aviation safety is perverse. A government minister wants to cut a ribbon on a new terminal. Nobody cuts a ribbon on an updated ATC training program.
There is one promising development. Aireon's space-based ADS-B network — this is a constellation of satellites that can track aircraft transponders from orbit, eliminating the radar coverage gap entirely. No ground stations needed. If an aircraft has ADS-B Out — which most modern airliners do — Aireon can see it anywhere on the planet, including over the DRC at five thousand feet.
The technology exists to fix the single biggest structural failure — the sixty percent radar gap below twenty-four thousand feet — without building a single ground radar station. What's the timeline?
The FAA and EASA are piloting it in African airspace now. Full deployment could happen by twenty twenty-eight if the funding and political will align. The issue is not the satellite network — that's already in orbit. The issue is getting national CAAs to adopt the data feed, train controllers on it, and integrate it into their procedures. That's a governance problem, not a technology problem.
Which brings us back to where we started. The opposite of aviation safety isn't bad airlines. It's broken systems. Yeti Airlines didn't crash because one pilot made one lever error. It crashed because a regulator that's been flagged for thirteen years allowed an airline with a history of hull losses to keep operating, with pilots who hadn't been trained to distinguish between a condition lever and a power lever. The individual error is the final link in a chain that was forged years earlier by institutional failure.
The DRC's eleven point two accident rate isn't caused by a single factor. It's the interaction of no radar, no approach aids, exhausted controllers, language barriers, and regulators with no enforcement capacity. You fix one and the others remain. You need systemic change, and systemic change is slow and expensive and politically thankless.
The answer to Daniel's question — which airline, which airspace — is Yeti Airlines and the Central African region, specifically the DRC. But the real answer is that the question itself reveals something about how we think about risk. We want a name. We want a carrier to avoid. And that's useful as far as it goes. But the name is just the surface expression of a regulatory and infrastructural collapse that's been decades in the making.
If you're booking a flight tomorrow, the practical advice is: check the blacklist, check the ASN database, check the USOAP scores, fly a carrier with IOSA certification and a modern fleet. But the bigger picture is that aviation safety is a public good, and public goods require functional institutions. Where the institutions fail, the accident rate follows — not immediately, but inexorably.
The airline is the symptom. The airspace is the disease.
And now: Hilbert's daily fun fact.
Hilbert: In the early fifteen hundreds, a Tuareg salt caravan crossing the Sahel was reportedly one wrong turn away from discovering a route that would have connected West African gold to the Pacific coast of South America — via the Atacama Desert — three decades before Pizarro reached Peru. The guide turned back because his camel sneezed.
...A camel sneeze changed the course of transcontinental trade.
I have so many questions about the geography of that route, and I'm going to suppress all of them.
This has been My Weird Prompts. If you got something out of this episode, rate us five stars and tell a friend who flies frequently — especially one who books tickets to places where the runway doesn't have lights.
We'll be back next week.
