The demand desk
What does India actually fly?
India's aviation regulator counts every passenger on every city pair and publishes it. Nobody puts those numbers next to what the route costs in points. Here they are — ranked by the people who actually fly them, with our cheapest verified award price beside each, and an honest blank where we haven't priced it.
how this is built, and what it can't tell you fold that back up
This is the demand. The award board is what it costs.
Directorate General of Civil Aviation, Government of India · domestic 12 months to May 2026, international 4 quarters to 2025-Q4 · read 2026-08-29 · the release
out of India · inside India · what we haven't priced · the method
Of the passengers DGCA counts flying out of India, 81% are on a corridor this desk has priced with a verified award chart. Inside India, that falls to 42%.
893
city pairs, ranked by passengers
234M
passengers behind them
253
of those pairs carry a verified award price
640
do not — the queue below
- These are city pairs, not airports. DGCA counts cities, so Delhi–Tokyo covers Haneda and Narita together and Delhi–Mumbai includes Navi Mumbai. We do not split a figure DGCA never split.
- The two boards cover different periods and must not be compared. Domestic is 12 months to May 2026; international is 4 quarters to 2025-Q4. DGCA publishes them on different cadences and different lags.
- A blank is a blank, not a zero. Where we have no verified price the cell says so. Where DGCA publishes no traffic for a pair, the pair is absent — it does not appear here as nobody flying it.
Out of India
The Gulf is not a rounding error on this list — it is the list. Dubai alone is one end of four of the ten busiest routes out of India, and the UAE of five. The long-haul corridors the points world writes about are the minority of what India actually flies.
4 quarters to 2025-Q4 · 370 city pairs · 77M passengers · 81% of them on a corridor we price with a verified chart
the next 30 busiest pairs
Inside India
This is where the desk is thinnest, and the number says so: 81% of India's international passengers fly a corridor we have priced with a verified chart, and only 42% of its domestic ones do. India flies a far wider domestic board than this desk prices.
12 months to May 2026 · 523 city pairs · 157M passengers · 42% of them on a corridor we price with a verified chart
the next 30 busiest pairs
What we haven't priced
The other direction of the same join, and the reason it is on a public page rather than in a desk note: these are the city pairs India flies in volume that carry no verified award price here. Busiest first. This is the order the next pricing wave works in.
Out of India
the 12 busiest pairs we can't price yet
Inside India
the 12 busiest pairs we can't price yet
Where these numbers come from
The source, and why we can cite it
The source of record is the Directorate General of Civil Aviation, Government of India, which publishes monthly domestic and quarterly international city-pair traffic as spreadsheets in an open bucket. The bucket refuses to list its contents, so filenames cannot be guessed — they are discovered from a maintained mirror, which also republishes the workbooks as parsed CSVs. We read the mirror and cite DGCA, and the reason that is honest rather than convenient is that the two were checked against each other: DGCA's own city-pair table for 2025-Q4 matched the mirror on 408 of 408 city pairs, with an identical passenger total of 2,08,30,121. That check is a command, not a claim, and it re-runs on every refresh.
DGCA's own table for 2025-Q4 · the mirror (ODbL 1.0, commit 306ff90852e0) · 525 primary files discovered
What a passenger count here means
Every figure is total passengers both ways across the whole window — DGCA's own to-and-from columns, summed over each month or quarter in it. It counts people, not flights and not seats. The windows lag: DGCA publishes domestic months in arrears and international a quarter behind, so nothing here is "today".
The join, and the trap inside it
DGCA names cities; this desk keys corridors on airport codes. Those are not the same thing, and the gap is not cosmetic. DGCA restyled its domestic city names in January 2026 — the row called MUMBAI stops, and one called MUMBAI (MUMBAI) starts. A rolling twelve-month window straddles that date, so adding up the raw names splits the busiest route in the country in two and reports Delhi–Mumbai at about 60% of its real size — in the right units, with nothing visibly wrong. So we map every published spelling, including DGCA's own misspellings, onto a metro, and aggregate there. Goa has appeared under five names; Bangkok and Don Mueang are reported separately, as are New York and Newark.
The consequence is stated on every row that has it: a figure covers all the airports we model for both cities. We map a city only to airports this desk already carries — we do not invent a code for an airport we have never plotted — so DGCA cities we cannot bind are dropped and counted rather than guessed at. That is 7% of domestic passengers and 1% of international ones, almost all of them small airfields with no award chart to price. The full list, with its passenger counts, is carried in the desk's own data file alongside every mapping above.
Licence and attribution
Contains data from the Directorate General of Civil Aviation (DGCA), Government of India, parsed by github.com/Vonter/india-aviation-traffic and made available under the Open Database License (ODbL) 1.0.
The data is DGCA's and the parse is the mirror's; this desk publishes the join and adds nothing to the numbers themselves. DGCA marks recent releases provisional.