America's Tightest Connections 2026: 43 of the 50 Busiest Airports Let an Airline Sell You a 30-Minute Layover

Two travelers book the same 30-minute layover at O’Hare. One is on a single ticket; the other stitched two cheap fares together. Same clock, same gates, same walk. If they both miss the second flight, the airline owes the first one a free seat on the next departure and owes the second one nothing at all. The layover was legal for both. Only one of them was protected.

I keep a database of the 50 busiest US airports and their published minimum connection times, one record per airport, each validated against the OAG minimum-connect-time database and date-stamped. Most of it feeds a free per-airport connection-time checker. This month I ran a single query across the whole set: what is the shortest connection an airline is legally allowed to sell at each of these airports, and how much of that legal window does the airport itself eat before you have even started walking.

Here is the finding, stated plainly. Of the 50 busiest US airports, 43 publish a 30-minute domestic-to-domestic minimum connection time. An airline can legally sell you a 30-minute connection at all 43. Only 7 airports set a higher floor, and the airports with the “worst” number on paper are the ones actually protecting you.

What a minimum connection time is, and how I built the number

A minimum connection time (MCT) is the shortest transfer an airline may legally sell as a single itinerary at a given airport. It is the number the booking systems enforce. If a connection falls below the airport’s MCT, the airline cannot sell it as one ticket. That is the entire definition, and it carries one crucial consequence: on a single ticket, a connection at or above the MCT is one the airline has deemed legal, so if you misconnect through no fault of your own, it must rebook you free. On two separate tickets you have no such protection, no matter how much time you left. The MCT measures legality, not advisability.

The figure I track is the OAG STANDARD MCT: the carrier-agnostic published floor for each airport, split into four sectors (domestic-to-domestic, domestic-to-international, international-to-domestic, international-to-international). I validated every airport’s four values against ExpertFlyer’s OAG MCT database on 2026-05-29, and the airport records were last verified on 2026-06-14. Where a dominant hub carrier files its own same-airline connection time that differs from the airport standard, I captured that too, and those exceptions are a story of their own below.

The full per-airport dataset, all 50 airport-standard rows plus the 26 same-airline carrier exceptions, is at /data/tightest-connections-2026.csv.

The domestic floor, from protective to bare minimum

The chart is the domestic-to-domestic MCT. Green is a protective floor above 30 minutes. Amber is a 30-minute floor at an airport where the terminals connect airside, so 30 minutes is at least a continuous walk. Red is a 30-minute floor at an airport where they do not, so the terminal change can eat the window. Bar width scales to LAX’s 70-minute floor.

Only 7 airports set a domestic floor above 30 minutes: Los Angeles at 70, Atlanta at 55, San Francisco at 50, Kansas City at 45, Philadelphia at 40, Cincinnati at 40, and Baltimore at 35. Every other airport in the top 50 publishes 30. That is the inversion worth sitting with. LAX has the “worst” connection number in the country, and it is the airport doing the most to keep you from booking a connection you cannot make. A 70-minute floor is a margin the airport built in. A 30-minute floor is that same margin handed to you to manage yourself.

The 17 airports where 30 minutes includes a terminal change

A 30-minute floor means something very different depending on whether you can reach your next gate without leaving security. At 17 of the 43 thirty-minute airports, you cannot. These airports are not airside-connected, so changing terminals means exiting the secure area, often riding a landside train or shuttle, and re-clearing security on the far side:

O’Hare (ORD), JFK, Orlando (MCO), Miami (MIA), Phoenix (PHX), Newark (EWR), Boston (BOS), Minneapolis-St. Paul (MSP), LaGuardia (LGA), Detroit (DTW), San Diego (SAN), Sacramento (SMF), Raleigh-Durham (RDU), Oakland (OAK), St. Louis (STL), Burbank (BUR), and Ontario (ONT).

Their own different-terminal transfer estimates run from 10 to 30 minutes. At ORD the ATS people mover between terminals is a 30-minute allowance. At JFK the AirTrain loop between terminals is another 30. At Newark the landside AirTrain monorail is 30. Boston’s landside Massport shuttle between terminals is 25. LaGuardia’s landside shuttle bus is 20, and Minneapolis-St. Paul’s landside light rail between its two terminals is 20. Line those numbers up against a 30-minute legal floor and the problem is obvious: at these airports, a legal 30-minute connection can be almost entirely consumed by the terminal change alone, before you account for a late inbound, a long taxi, or a single stop for a bathroom. The number is legal. It is not always possible.

What the hub carriers know

The most honest numbers in the dataset are not the airport standards. They are the times the airlines file for their own same-airline connections at the airports they know best. There are 26 of these carrier exceptions across the 50 airports, and they split into a clean pattern: 7 cut below the airport standard, 12 raise above it, and 7 match.

The cuts all happen at fortress hubs where the carrier controls both flights and can hold a plane or rebook you internally, and where an airside people-mover makes a fast transfer physically real:

• Delta at Atlanta: 40 minutes, against a 55-minute airport standard
• American at LAX: 35, and United at LAX: 30, and Delta at LAX: 40, all against LAX’s 70-minute standard
• American at Phoenix: 25, against the 30-minute standard
• United at San Francisco: 35, against 50
• American at Philadelphia: 30, against 40

The raises are just as telling, and they cluster at the congested, shared, terminal-split airports where the airline that flies there most gives itself more room than the airport legally requires, all against a 30-minute standard:

• American at Dallas-Fort Worth: 40
• United at Denver: 40
• United at O’Hare: 35, and American at O’Hare: 35
• Delta at JFK: 45, and JetBlue at JFK: 45
• American at Miami: 40
• Alaska at Seattle: 40
• United at Newark: 45
• United at Houston (IAH): 40
• JetBlue at Boston: 45
• Delta at LaGuardia: 45

Read those two lists together and the signal is unmistakable. Where a hub carrier controls the airport and the transfer is genuinely fast, it cuts below the standard. Where the airport is congested and split across terminals, the carrier that operates the most flights there raises its own floor above what the airport requires. When the airline that knows an airport best gives itself more time than the law demands, that raised number is the real minimum, not the 30 on the airport’s sheet.

And the single lowest legal number anywhere in the dataset is that Phoenix line: American Airlines same-airline connections at 25 minutes. It is legal. Whether 25 minutes is ever a good idea is a question the MCT does not answer.

International arrivals: the 90-minute wall

The moment an international flight is involved, the floor jumps. At 46 of the 50 airports the published international-to-domestic MCT is exactly 90 minutes. That single window has to absorb taxi-in, immigration, baggage reclaim, customs, re-checking the bag onto the domestic leg, and re-clearing TSA. Ninety minutes is legal, and on a heavy arrivals bank it is genuinely tight.

Four airports set it higher because they know the arrivals crush is worse: LAX at 120 minutes, SFO at 105, BWI at 75, and CVG at 60. LAX again sits at the protective extreme. A two-hour international-to-domestic floor is not the airport being slow; it is the airport being honest about how long its own arrivals process takes.

What this study does not capture

I want the limits on the table as clearly as the findings.

This is a measure of legality, not advice. Every number here is what an airline may sell, not what will comfortably work. A connection can clear the MCT and still be a bad bet on a day with weather, a late inbound, or a packed security line.

I have not crossed this data with delay or queue data yet. The MCT is a static published floor. It does not know that your inbound is chronically 20 minutes late, or that the TSA line at that terminal spikes at 6pm. Layering historical delay and wait-time data over these floors is the obvious next build, and I do not have it structured yet.

The carrier-exception list covers the dominant hub carriers where OAG captured a same-airline exception. It is not every airline at every airport. A carrier that simply uses the airport standard has no exception row, which is correct, but it means the 26 exceptions are the notable deviations, not an exhaustive airline-by-airline grid.

And this study is US-only, the 50 busiest US airports. The same database holds international airports, but non-US airports are out of scope here.

What to actually do with this

Three practical moves fall out of the data.

Book the hub carrier’s own metal at its hub. The raised carrier exceptions exist because the airline that runs the airport knows how long a transfer really takes there. If you are connecting at Newark, a United-on-United itinerary is priced against United’s own 45-minute floor, not the airport’s optimistic 30. That number is the airline betting its own rebooking costs on the transfer, which makes it the number to trust.

Treat one ticket and two tickets as different products. A 30-minute connection on a single itinerary is a legal connection the airline must protect. The same 30 minutes across two separate tickets is a gamble with no backstop. If you are self-connecting on separate tickets, ignore the MCT entirely and give yourself the kind of margin the airport’s protective floors imply, closer to LAX’s 70 than to the 30-minute minimum.

Prefer airside-connected airports for genuinely tight transfers. If the itinerary forces a short connection, an airport where you never leave security (Atlanta, Charlotte, Denver, Dallas-Fort Worth) is a categorically safer place to do it than one where 30 minutes might be spent on a landside shuttle and a second TSA line. You can check any specific airport against its own transfer setup with the connection-time checker, which reads the same date-stamped database as the numbers above.