How these wait times are worked out

Two different kinds of number appear on this site, and we label which is which on every page.

Border crossings: measured

Every border number on this site is CBP's own. U.S. Customs and Border Protection publishes a live feed of delays at all 80 land crossings it staffs, broken out by passenger vehicles, pedestrians and commercial trucks, and again by general, Ready and trusted traveller lanes. We read that feed directly in your browser, so what you see is what CBP is publishing at that moment, not a copy we cached this morning.

When a lane shows no reading, CBP is not reporting on it. That usually means the lane is closed. We show the gap rather than filling it with a plausible number.

Airport security: the shape is measured, the minutes are modelled

There is no national feed of TSA checkpoint waits. Anyone who shows you a live security wait for every US airport is showing you an estimate, whether they say so or not. We would rather say so, and then explain the estimate.

What is measured. The Bureau of Transportation Statistics publishes every scheduled departure by every reporting airline: the airport, the local departure time, the date. We use three months of it, more than 1.8 million departures, to build the hour-by-hour departure load at each of the 355 airports on this site. That is the real driver of a checkpoint queue, and it is a fact, not a guess. It is what the shaded grid on every airport page shows.

What is modelled. Turning that load into minutes takes three assumptions, and here they are in full:

• Passengers reach the checkpoint ahead of their own departure, spread across the preceding hours rather than all at once. We weight a flight's contribution to a given hour's queue at 10 percent for the hour it departs, 45 percent one hour earlier, 32 percent two hours earlier and 13 percent three hours earlier.
• An airport's checkpoint capacity scales with the traffic it was built for, so what matters is how hard an airport is pushing against its own peak, not its size in absolute terms. A large hub at full stretch has a wider spread between its quietest and busiest hour than a regional field does.
• Queues grow faster than load once a checkpoint approaches capacity, so the curve is bent rather than straight.

Because the minutes are modelled, we publish a band rather than a single number. A page that tells you the wait is exactly 23 minutes is telling you something it cannot know.

The three lanes. Standard is the number above. PreCheck runs roughly a third of it, floored at a few minutes because no lane is instant and capped so it never claims an implausibly short wait at a jammed checkpoint. CLEAR is shorter still: it is not a lighter screening but a queue position, so we model it just below PreCheck and never at zero, because you still walk through the same scanner behind whoever CLEAR placed ahead of you.

Terminal walk times: real gate positions, modelled minutes

On the terminal-map pages we show how long it takes to walk from the security checkpoint to each gate. The geometry is real: every gate is an actual coordinate from OpenStreetMap, whose aeroway=gate nodes cover the major US airports gate for gate. The one thing no dataset publishes is where the checkpoint itself is, so we place that by hand, one point per terminal, and mark each one unverified until it is checked against the airport's own map.

The minutes are modelled, the same way the security waits are. We take the straight-line distance from the checkpoint that serves a gate, multiply by a corridor factor of 1.3 because nobody walks through walls, and divide by a measured free-flow walking speed of 264 feet per minute, with a standard deviation of 52 feet per minute, from Airport Cooperative Research Program (ACRP) Report 67. Because both the path and the walker vary, the answer is a band rather than a single number. It does not include moving walkways, escalators or a train between concourses: it is the walk, on foot. We publish a terminal page only where the gate coverage is real and a checkpoint has been placed, because without the walk minutes there is nothing on the page a plain map does not already have.

Ferries: a published timetable and a modelled boarding buffer

The ferry pages read the Washington State Ferries schedule and, where the feed carries them, vessel positions and per-sailing car-deck space. The schedule is a published timetable, not a model. Vessel positions carry the operator's own observation time and are baked at each rebuild, not tracked moment by moment, and the page says so; the crossing duration is shown only where WSF actually publishes it, and left blank where it does not.

The one modelled number is the boarding buffer: to turn a sailing time into a leave-by we assume you should reach the tollbooth 20 minutes before departure, which is WSF's own advice for a drive-up vehicle. So leave home = sailing time − your drive to the dock − 20 minutes. Change the drive to your own and the answer moves with it; the buffer is the single assumption, stated on the page and here.

What we will not do

We do not publish checkpoint opening hours we cannot source, we do not invent live numbers for airports that publish none, and we do not quietly turn a model into a measurement. Where an airport runs its own official live wait feed, that number wins and the model steps aside; those are being added one airport at a time.

Sources

• U.S. Customs and Border Protection, border wait times feed.
• U.S. Bureau of Transportation Statistics, Reporting Carrier On-Time Performance, May 2026, July 2025 and December 2025.
• OurAirports and OpenFlights for airport names, locations and time zones.