Does this tell me whether I will make my connection?

No, and nothing could. It shows how often arrivals into this airport landed inside a given cushion over the three years in the window. Taxi time, the walk between concourses, how long the airline holds a connecting flight, and how many people are ahead of you at security are all outside these filings.

Which flights are counted?

Every scheduled arrival at the airport in the window. A cancelled flight never arrived; a diverted one reached the airport hours off schedule if it reached it at all. Both are kept out of the lateness shares and reported on their own line — treating either as an extremely late arrival would flatter nobody.

Why do the cushions jump from 60 to 90 to 120 minutes?

Because those are the cut points the filings are broken at. A layover that lands between two cuts is read at the lower one: a 75-minute cushion is reported at 60 minutes, which understates the odds rather than flattering them.

Is a legal connection the same as a safe one?

No. Airlines publish a minimum connection time for each airport — often 35 to 45 minutes domestically — and that is the shortest connection they are willing to sell, not the one that usually works. The numbers here describe how the inbound flights actually landed, which is the part the schedule cannot promise.

Connection Risk — What a Layover Had to Absorb

Name: US airport arrival lateness distribution
Creator: Bureau of Transportation Statistics
License: https://www.usa.gov/government-works

A layover is a bet on the inbound flight. Here is how that bet has run: for every arrival into an airport over the three years through June 2026, how far past schedule it actually landed — and how often that was inside the cushion you have. These are historical odds, not forecasts. The records say what happened on this schedule over three years; they know nothing about tomorrow’s thunderstorm.

Data through June 2026 · Bureau of Transportation Statistics

Airport you connect through

Layover, in minutes

The pickers need JavaScript to redo the arithmetic. What is shown below is ATL with a 60-minute cushion; every airport has its own page carrying its full record, starting from the airport index.

Hartsfield Jackson Atlanta ATL

Over the three years through June 2026, 93.7% of arrivals into ATL landed no more than 60 minutes past schedule. A 60-minute layover would have had that much room for the inbound flight alone.

Of the arrivals that were late at all, 63.6% were no more than 60 minutes late — 106,948 of 168,280.

Separately, 1.6% of scheduled arrivals into ATL were cancelled or diverted off their route — either way not an arrival anyone could have planned a connection around. That is 15,275 of 984,098 scheduled.

Every cushion these filings can answer

15 min 82.6%
30 min 88.7%
45 min 91.8%
60 min 93.7%
90 min 95.9%
120 min 97.1%

Share of arrivals that landed no more than that many minutes past schedule. The cushion you entered is marked.

How the arrivals landed

Every scheduled arrival that landed, by how late it was.
Landed	Share of arrivals	Flights
Landed less than 15 minutes behind schedule	82.6%	800,543
15 to 30 minutes late	6.1%	59,217
31 to 45 minutes late	3%	29,207
46 to 60 minutes late	1.9%	18,524
61 to 90 minutes late	2.2%	21,601
91 to 120 minutes late	1.2%	12,072
More than two hours late	2.9%	27,659

Everything else about ATL — hours, months, airlines

How to read this without kidding yourself

These are arrivals that already happened, not a forecast and not advice. The share tells you how often the inbound flight would have left a given cushion intact — nothing more. The connection itself also depends on the gate you land at, the concourse you have to cross, the airline's own rules about holding a flight, and whether your bag makes the same trip.

Cancelled and diverted flights are counted separately, because a flight that was cancelled or sent somewhere else is not a late flight — it is a different problem, and rolling it into the lateness shares would quietly make every airport look worse at one end and better at the other.

The other tools

Best time to fly — Pick an airport and read the day: what share of departures left on time in each two-hour band, Monday through Sunday.
Route reliability — An origin and a destination, then the route’s record airline by airline — how often flights arrived on time, how late they ran when they did not, and the months to avoid.
Airline comparison — Every reporting carrier side by side: arrival on-time rate, cancellation rate, and what the delay minutes were put down to.

Questions travelers ask

Does this tell me whether I will make my connection?: No, and nothing could. It shows how often arrivals into this airport landed inside a given cushion over the three years in the window. Taxi time, the walk between concourses, how long the airline holds a connecting flight, and how many people are ahead of you at security are all outside these filings.
Which flights are counted?: Every scheduled arrival at the airport in the window. A cancelled flight never arrived; a diverted one reached the airport hours off schedule if it reached it at all. Both are kept out of the lateness shares and reported on their own line — treating either as an extremely late arrival would flatter nobody.
Why do the cushions jump from 60 to 90 to 120 minutes?: Because those are the cut points the filings are broken at. A layover that lands between two cuts is read at the lower one: a 75-minute cushion is reported at 60 minutes, which understates the odds rather than flattering them.
Is a legal connection the same as a safe one?: No. Airlines publish a minimum connection time for each airport — often 35 to 45 minutes domestically — and that is the shortest connection they are willing to sell, not the one that usually works. The numbers here describe how the inbound flights actually landed, which is the part the schedule cannot promise.