Every reassurance on this site traces back to this one figure. Here is what it is, what it categorically is not, and how to link to it instead of re-typing it.
Between 2018 and 2022, one passenger died for every 13,700,000 passenger boardings on scheduled commercial flights, worldwide. A boarding is one passenger taking one flight — not one departure of an aircraft carrying hundreds of people. That is the entire derivation of 99.999993%: one minus one divided by 13,700,000.
MIT / ARNOLD BARNETT, 2024 → (opens in a new window)0.000007% is the chance any one boarding ends in a passenger death — one divided by 13,700,000, as a percentage. Subtract it from 100% and the remainder is 99.999993%, the figure printed everywhere else on this site. Nothing here is rounded before the last step.
A single flight carries hundreds of boardings, one for each seat filled. Dividing by flights instead would understate the risk many times over — this site publishes only the boarding figure.
It counts passenger deaths, not accidents. Most accidents in the underlying data killed nobody at all — a death rate and an accident rate answer different questions.
The International Air Transport Association publishes its own annual figures, different methodology, different number. This one is Barnett and Reig Torra's — a mix-up this site itself shipped once, now blocked by an automated check.
The same paper found the rate has roughly halved every decade since the 1960s: commercial flying is not just safe today, it keeps getting safer, decade over decade.
Illustrative only — no per-decade rate is published, so none is plotted. The claim is the source's, not a chart of real values.
MIT / ARNOLD BARNETT, 2024 → (opens in a new window)Full reference: Barnett & Reig Torra, Journal of Air Transport Management, 2024 — worldwide passenger boardings, 2018–2022.
This figure set against everyday risks, row by row: /odds. How every number on the site is sourced: /methodology. The real, dated events behind the record: /the-record.