Flight Cancellations Hit a Four-Year High. Book Morning.

The useful number this week is not a new airport ranking for its own sake. A Sept. 30 Stacker piece carried by Silicon Valley says U.S. carriers have canceled roughly 2% of flights since Jan. 1, according to FlightAware, and that this is the highest cancellation rate since 2022. The same piece says summer cancellations on U.S. carriers were about 15% higher than last summer. I am using those two figures as the reason to care which airport you pick, and which hour you pick. A list of bad airports without a year like this one is just a list.

We already wrote the policy side of a ruined trip in what travel insurance actually pays for. We wrote the satisfaction survey in MSP winning again, and O’Hare tying Newark at the bottom. We wrote the ticket-error problem in what to do when an airline cancels your reservation by mistake. This is the operational year those pieces sit inside. The Stacker writeup does not name one cause. That absence is part of the story.

Flight cancellations are a year rate, not a storm story

FlightAware’s year-to-date figure, as Stacker prints it, is about 2% of U.S. carrier flights canceled since Jan. 1. Two percent does not sound like a crisis until you put it next to 2022, which the piece calls the last time the rate was this high. 2022 was the year after the worst of the pandemic staffing mess, when one airline after another melted down in public. Stacker is not saying 2026 is that year again. It is saying the cancellation rate has climbed back to that neighborhood, and that the summer slice was about 15% worse than the summer before it.

The piece is also clear that delays did not jump the same way. About one in four departures was delayed nationwide between Memorial Day weekend and Labor Day, and that delay rate did not move meaningfully from last year. So the year is worse at the hard failure, the canceled flight, and roughly the same at the softer one, the late departure. If you are choosing a trip, those are different problems. A delay often still gets you there the same day. A cancellation is a new ticket, a hotel, or a night in the terminal. The 15% summer jump is the one I would plan around.

On Sept. 21, the piece says, airlines shuttered more than 1,000 departures after an air traffic control equipment outage snarled the East Coast. Operations resumed a day later. Stacker points at The Points Guy for more on that day. I did not read that follow-up, so I am not going to borrow extra claims from it. What I can use is the scale: more than 1,000 departures in one day, on top of a year that was already running hot. One equipment outage does not explain a 2% year. The piece says there seems to be no single problem behind this year’s disruptions. I would rather keep that sentence than invent a villain. Weather, staffing, a bad afternoon bank, and a piece of failed equipment can all show up in the same percentage.

If you only take one planning move from the year rate, stop treating a cancellation as a freak event you will argue your way out of at the counter. Two percent is still a small share of flights. It is large enough that a two-leg trip has more than one chance to hit it, and a family of four on separate tickets has more chances still. The insurance guide we already published is about what a policy pays after you are stuck. This week’s numbers are about lowering the chance you need the policy.

New York owns the cancellation list, and Newark is the least bad of the three

Cirium’s cut, as Stacker reports it, is U.S. airports with at least 30,000 flights between Memorial Day and Labor Day. That floor matters. A tiny airport with one canceled regional flight can post an ugly percentage and mean nothing for your trip. These lists are large airports.

The worst cancellation rates in that window:

1. LaGuardia, 6.5% of flights canceled.
2. Reagan National, 5.8%.
3. Kennedy, 4.2%.
4. Boston Logan, 3.8%.
5. Newark, 3.7%.

Three of the five are the New York area’s main hubs. Stacker says that outright: if you are flying out of the New York area, the data shows you are more likely to have a flight canceled than at other airports. All three of the region’s busiest hubs landed on the national worst list. That is the line I would put on a fridge if someone in the house is booking a fall connection through the city.

The follow-up question in the piece is which of those three is the least bad bet. After a rough 2025, Newark was the New York area’s best airport for delays and cancellations this summer, still according to Cirium via Stacker. “Best” here is a relative word. Newark’s cancellation rate on this list is 3.7%, which is still on the national worst-five board. It beat LaGuardia and Kennedy. It did not beat Salt Lake City, and it did not beat the national year rate of about 2%. If you have to connect in the New York area, Newark is the number the summer data prefers. If you can avoid the area, the list is telling you to avoid it.

Boston at 3.8% is close enough to Newark that I would not rebuild a trip to swap them. Reagan at 5.8% and LaGuardia at 6.5% are a different tier. A 6.5% cancellation rate means roughly one flight in fifteen did not operate in a window that already included the busy summer. That is not a reason to swear off LaGuardia forever. It is a reason not to book the last flight of the day out of it if you have a wedding, a cruise, or a meeting you cannot miss.

Our J.D. Power piece had a different Newark story: satisfaction, not operations, and O’Hare tied with Newark at the bottom of that survey. Both can be true. A passenger can hate the building and still have a better chance of the plane leaving than they would at LaGuardia. Do not mix the two rankings when you pick a ticket. Satisfaction is how the airport felt. Cirium’s percentage is whether the flight operated.

The delay list is not the cancellation list

The airports that delayed people are mostly not the airports that canceled them. Stacker’s delay ranking, same Cirium window, same 30,000-flight floor:

1. Nashville, 35.3% of flights delayed.
2. Baltimore/Washington, 34.6%.
3. San Diego, 33.6%.
4. Dallas/Fort Worth, 33.4%.
5. San Francisco, 32.1%.

About a third of departures late, at airports that do not appear on the cancellation worst-five at all. Nashville’s problem, on these numbers, is lateness, not disappearance. San Francisco’s is the same. If your trip can absorb an hour and cannot absorb a canceled flight, those airports are a different risk from LaGuardia. If your trip is a tight connection plus a same-night event, a 33% delay rate is its own kind of cancellation. You just do not get the refund conversation. You get the missed connection.

I would not treat 35.3% and 32.1% as a meaningful gap for a booking decision. Both are “about one in three.” The national figure in the same piece is about one in four departures delayed. The worst delay airports are worse than the country, not a different species of airport. What I would treat as meaningful is the split between the two lists. New York and Boston and Reagan are where flights vanished. Nashville, Baltimore, San Diego, Dallas, and San Francisco are where flights slid. Plan the buffer for the failure mode the airport actually had.

Stacker says delays did not jump meaningfully from last year even while cancellations did. So a delay ranking that looks familiar is not a comfort. It is the baseline. The new pain is the cancellation column. If you flew through Dallas last summer and remember being late, this summer’s data says that part did not get dramatically worse. The part that got worse is the flight that never left, and that part concentrated in a short list of Northeast airports.

The on-time hubs are connection airports, if you can use them

The best on-time large airports in the same Cirium cut:

1. Salt Lake City, 82% of flights on time.
2. Minneapolis-St. Paul, 80.1%.
3. Dulles, 79.4%.
4. Los Angeles, 78%.
5. Houston Intercontinental, 77.5%.

Stacker frames these as hubs to aim for in summer 2027, especially if you have to connect. I would use them this fall too, with one caveat. An on-time rate is not a promise about your specific afternoon. It is a summer average for airports that flew at least 30,000 times. Salt Lake City at 82% still means almost one flight in five was not on time. Minneapolis at 80.1% is the airport that also topped the satisfaction survey we wrote up. Two different measurements, same airport, both favorable. That is as close as this week gets to a cross-check, and it is still not a reason to book a 40-minute connection.

Dulles at 79.4% is the interesting one if you are stuck in the Washington area. Reagan is on the cancellation worst list at 5.8%. Dulles is on the on-time best list. They are not interchangeable airports, and the cab ride between them is not free, but if the trip is a connection rather than a downtown meeting, the summer numbers prefer Dulles by a wide margin. I would not make that swap for a 7 a.m. flight you can reach easily from the city. I would make it for an evening connection you cannot afford to lose.

Houston and Los Angeles are on this list because the percentage says so, not because they are pleasant. An on-time hub can still be a long walk and a bad food court. Use the percentage for the connection. Use the satisfaction survey, if you have one, for the hours you will spend in the building. Do not ask one number to do both jobs.

Book the morning flight, then stop optimizing

The tips in the Stacker piece are less glamorous than the rankings, and they are the part I would actually follow.

Book the morning flight. Morning flights tend to run more smoothly than afternoon and evening flights, because the day’s problems compound. Maintenance issues show up. Afternoon storms fire up. Planes fall behind as they cross the country. The quote Stacker borrows from The Points Guy is Frontier chief executive Jimmy Dempsey: “If you start cleanly in the morning, the airline tends to run on time for the rest of the day.” I have not watched the interview. I am using the sentence the Sept. 30 piece printed, and it matches the mechanism the piece describes. The first flight of the day has not inherited a late inbound. The 7 p.m. flight often has.

That is a harder tip than it sounds if you hate early alarms. It is also cheaper than buying a backup ticket. If the event you are flying to cannot move, the morning departure is the buffer you can buy without paying for a second seat. If you can only get an evening flight, the same logic says you should not connect. A late departure plus a connection is two chances for the day’s mess to land on you.

The piece is specific about connections: be especially cautious about an itinerary with a connection in the late afternoon or evening, particularly during peak spring and summer storm season. Fall is not that season, and I am not going to pretend October thunder is July thunder. The compounding problem does not require a thunderstorm. A late aircraft is enough. I would still rather connect before noon in October than at 6 p.m., even if the storm clause in the tip is aimed at summer. The Dempsey sentence is not a summer sentence. It is a description of how an airline’s day works.

A credit card with travel protections is the other tip, and it has a condition people skip. If weather strands you, the airline typically will not cover meals or a hotel. The card might. The claim usually requires that you booked the trip on that card. Buying the ticket on a debit card and hoping the travel card in your wallet will cover the hotel is how those claims die. We walked through what a standalone policy covers, and what it skips, in the insurance guide. The card benefit is a narrower tool. It is still the tool the Sept. 30 piece says matters more now, because starting in October airlines may not cover your costs for some mechanical delays.

I want to be careful with that last line. Stacker states it. The extract I could read does not name the rule, the docket, or which delays count as “some.” I am not going to invent a Department of Transportation citation to make the sentence sound harder. What I will do is treat October as a month when you should not assume the airline will pay for the hotel if the plane breaks. Book the trip on the card that has the protection. Keep the receipt. Read the benefit guide before you need it, not in the terminal.

The piece also tells you to check airline apps for flight updates. I do not have the rest of that section, so I am not going to pad it with generic app advice. The morning flight, the connection hour, and the card you booked on are the three moves the text actually supports.

What these lists will not do for you

Cirium’s floor of 30,000 flights keeps small airports off the board. If you are flying a regional airport that did not make the cut, you do not get a free pass and you do not get a warning. You get silence. FlightAware’s 2% is a national carrier rate, not your airport’s rate. Do not average them in your head and call it a personal forecast.

The lists also will not tell you which airline inside the airport had the bad summer. An airport rate mixes every carrier that used the field. A 6.5% cancellation rate at LaGuardia can be one airline’s problem or everyone’s. Stacker does not split it. If you have a choice of carriers on the same route, this article cannot pick the carrier for you. It can tell you the field was rough.

I would not rebuild a paid ticket because an airport sits one place higher or lower on a five-name list. I would rebuild a ticket that connects through LaGuardia at 8 p.m. when a morning nonstop exists. I would rebuild a ticket bought on a debit card if the trip is nonrefundable and the only protection you have is a card benefit you failed to trigger. Those are the decisions the numbers change. The rest is a ranking you can read and then go book the early flight.