The Department for Transport publishes a worked example of a London to Glasgow journey, and it is more useful than any of the campaigning material written about the same trip. Per passenger mile, it puts a train at 0.06 kilograms of carbon dioxide equivalent in direct emissions and 0.01 in indirect. It puts a plane at 0.26 direct and 0.05 indirect. A coach comes in below both, at 0.04 and 0.01.
Those are the headline numbers and they are not the interesting part.
What the footnotes do to the comparison
The same document attaches two conditions to the plane figure. The first is that it excludes a further 62 kilograms of carbon dioxide equivalent for indirect effects across the journey. The second is that it excludes 6 kilograms for getting to the airport by car.
Both additions are large next to per-mile figures measured in hundredths of a kilogram, and neither has an equivalent on the rail side of the comparison, because a train station in a British city is generally somewhere a person can already get to. That asymmetry is doing more work in the final comparison than the direct emissions are, and it is the part that gets dropped when the numbers are repeated elsewhere.
The indirect effects in question are the ones aviation has at altitude and surface transport does not: contrail formation and the atmospheric chemistry that follows it. The Department's guidance on aviation emissions methodology describes an uplift applied on top of direct emissions to account for them. How large that uplift should be is genuinely contested in the literature. That it belongs in the comparison at all is not.
The car is the actual competitor
Framing this as trains against planes flatters both. The National Travel Survey puts car travel, as driver and passenger combined, at 59 per cent of all trips and 76 per cent of all distance travelled in England in 2024. Nothing else is close.
The Department's worked example prices a petrol car at 0.17 kilograms direct and 0.05 indirect per passenger mile, and a diesel car at 0.18 and 0.04. Read quickly, that looks like a car is roughly two thirds of a plane and three times a train.
Read carefully, it is not, and the document says why: its car conversion factors assume a single passenger, and it adjusts its own worked example for an average occupancy of 1.5. A car carrying four people on the same route is a different proposition from a car carrying one, and the per-passenger figure moves accordingly. A train's figure does not move for the reader, because the occupancy assumption is already baked into it and is not the passenger's to change.
This is why the honest version of the advice is narrower than the campaigning version. For one person travelling alone between two British cities that have a direct rail service, the train wins on emissions by a margin large enough that the uncertainties do not threaten it. For a family of four driving to somewhere the railway does not go, the arithmetic is a different arithmetic, and telling them otherwise is how a reader learns to stop trusting the source.
Where the substitution actually works
The geography is the constraint nobody publishes a conversion factor for.
The domestic flights with meaningful passenger numbers are the ones where rail is slow relative to flying, not the ones where it is fast. London to Edinburgh and London to Glasgow are the routes where the argument gets made most often, and they are also the routes where rail is most competitive on door to door time once airport processing is included. The routes where flying genuinely saves hours, from the south of England to the far north of Scotland, or anything involving a sea crossing, are the ones where the emissions case runs into a wall that is not about emissions.
None of that is a reason to fly London to Manchester. It is a reason to stop treating "take the train" as advice that applies uniformly, because a reader who tries it on the wrong route finds out that it does not and discounts the rest.
How much of this there is to substitute
Two numbers set the size of the question, and they are worth putting next to each other even though they do not measure quite the same thing.
The Civil Aviation Authority recorded 295 million passengers through UK airports in 2024, of which 11.6 million were on domestic flights to or from London's six major airports. The Office of Rail and Road recorded 1,729 million passenger rail journeys in the year to March 2025, or 1,486 million excluding the Elizabeth Line, up 7.2 per cent on the year before.
The comparison is imperfect and the imperfection matters. The aviation figure is a calendar year and counts passengers through airports. The rail figure is a financial year and counts journeys. The domestic aviation number covers London airports rather than the whole country, so the true domestic total is higher. None of that is enough to close a gap of two orders of magnitude.
What the gap means is that domestic aviation is not where most British journeys are, and substituting all of it would move the national total by less than the headlines about it imply. It is a visible, tractable, symbolically large target rather than a numerically dominant one. That is a reasonable thing for a campaign to pick. It is not the same as it being the biggest available saving, and the two get conflated constantly.
The price problem, which is not a footnote
This is where the published data runs out and it is worth being explicit about that rather than filling the gap.
Anyone who has priced a walk-up rail fare against an advance-booked flight on the same route knows the direction of the result. What is harder to establish from a single citable source is the size and consistency of that gap, because rail fares in Britain vary by booking window, time of day, railcard, split ticketing and operator in ways that resist a single number. A figure asserting that trains cost some specific multiple of flights would be easy to write and impossible to stand behind.
What can be said is what the structure does. Air fares on these routes are set by yield management from a low base. Regulated rail fares are not, and the unregulated ones behave differently again. The result is that the environmentally better option is frequently the more expensive one at the point of decision, and no amount of per-mile carbon accounting changes what somebody does when they are looking at two prices.
What the numbers are for
The conversion factors exist because organisations have to report emissions, not because they were designed to settle an individual's travel decision. Using them for the second purpose is reasonable, and it comes with the caveats above: the occupancy assumption, the indirect effects that sit outside the headline figure, and the airport transfer that is real but rarely counted.
Applied honestly, they support a claim narrower than the one usually made and considerably more durable. Between British cities with a direct service, for a person travelling alone or in a pair, rail beats flying by a margin that survives every reasonable adjustment to the assumptions. Outside those conditions the answer depends on the conditions, and a publication that says so is more use than one that does not.