A benchmark is a reference number you compare your own against: what a seat-kilometre costs, how much an aircraft flies, the load factor, the fuel per passenger. Everyone uses them — airlines negotiating with lessors and MRO shops, investors, regulators, and we in the calculator when we check our model. But almost every benchmark comes with small print, and a head-on comparison more often shows differences in networks, accounting and currencies than in efficiency. Below: which benchmarks the industry has, why they are built the way they are, what their pitfalls are and when to use which. And at the end, our own airline efficiency rating from public 2025 reports, built to avoid those pitfalls as far as open data allows. We will update it every year.
Almost all benchmarks rest on a few unit metrics. An available seat-kilometre (ASK) is one seat flown one kilometre, occupied or not; it measures supply. A revenue passenger-kilometre (RPK) is one passenger carried one kilometre; it measures sales. Their ratio is the load factor: the share of seats offered that were sold. Cost per ASK (CASK) is all operating expenses divided by ASK; revenue per ASK (RASK) is the same for revenue. RASK minus CASK is the operating profit on every seat offered. Next to them sit physical metrics: utilisation — block hours per aircraft per day — and fuel burn per passenger-kilometre.
The appeal of these metrics is that they remove scale: an airline with 40 aircraft can be compared with one with 900. Their weakness is that they remove nothing else: where the airline flies, on which aircraft, under which accounting rules and in which currency. McKinsey says plainly that CASK does not show where the levers for cutting costs actually are2, and IATA admits in its maintenance cost report that cost benchmarking is “not a science” and that no normalisation exists that would allow direct comparisons1. That is why a whole ecosystem of methods, adjustments and arguments has grown around these simple formulas.
The most obvious source is the airlines' own reports. They are written for investors, and each airline picks the metric that suits its business model: Ryanair reports cost per passenger19, easyJet per seat5, US carriers per seat-mile. Reports are open and current, but definitions differ, so they cannot be compared directly.
The richest open dataset is the US regulatory filing Form 41. Every US carrier with revenue of $20 million or more files an income statement, aircraft costs by type, fuel and traffic with the Department of Transportation7,20,21. A uniform chart of accounts makes the costs comparable by definition — the only such dataset in the world, and the one our calculator's calibrations rest on. The MIT Airline Data Project turns these forms into ready-made series3. There is one weakness, but a big one: it covers only the US.
IATA's industry programmes work on a “give data, get a benchmark” basis. The maintenance cost programme (MCX) is free for member airlines that submit data every year; its public 2024 edition covers 28 airlines and 2,703 aircraft, about 8% of the world fleet, and gives an average of $1,522 per flight hour1,22. Anonymity is the condition without which airlines do not share their costs. Consultants mostly repackage the same Form 414, and next to cost benchmarks stand quality benchmarks: punctuality, delays, passenger satisfaction13,16,23.
| Benchmark | What it measures | Coverage | Access | When to use |
|---|---|---|---|---|
| Airline reports | annual reports, 10-K and 20-F, results releases | all listed carriers, each with its own definitions | open | primary source for the rating and for checking the model5,6 |
| US DOT Form 41 and T-2 | income statement, costs by aircraft type, fuel, traffic | all US carriers with revenue of $20 million or more, a uniform chart of accounts | open, free | block-hour cost by type, utilisation, fuel burn7,8 |
| MIT Airline Data Project | metric series based on Form 41 with a stage length adjustment | about 15 US carriers | open | reference methodology, historical series9 |
| IATA MCX | maintenance costs per flight hour, cycle and aircraft | 28 airlines, 2,703 aircraft (2024) | full for participants, public version open | negotiating with MRO shops and lessors1 |
| IATA ACMG | cost comparison by aircraft type among participants | programme participants | by participation | an airline's internal benchmark10,11 |
| ICAO, EUROCONTROL, FAA | state financial statistics, cost of delay, block-hour cost by aircraft class | world, Europe, US | mostly open | project appraisals, cost of delay12,13 |
| National regulators (UK CAA, Rosaviatsia) | traffic, load factor, utilisation by type for UK carriers | domestic carriers | open | utilisation and load factor; no financials14,15 |
| Consultants (Oliver Wyman, McKinsey, Aviation Week, CAPA) | CASK comparisons, block-hour cost | varies; for the US based on Form 41 | articles partly open, databases by subscription | ready-made comparisons, methodology in footnotes4 |
| Punctuality and quality (Cirium, CODA, Skytrax) | on-time share, delays, passenger surveys | world, Europe | Cirium monthly reports free | quality next to costs — but not in the index16 |
| Fuel efficiency (ICCT, sustainability reports) | passenger-km per litre, CO₂ per passenger-km | selective; ICCT's latest data are for 2017 | open | the fuel pillar of the rating17,18 |
Russia has no open equivalent of Form 41: state statistics give passengers, passenger-kilometres and load factor, but not seat-kilometres by carrier, utilisation or financials24. So the only Russian carriers in our rating are those that publish detailed reports themselves.
Every flight carries costs that do not depend on distance: take-off and landing, airport charges, ground handling, time on the ground. On a short flight they are spread over few seat-kilometres, on a long one over many25. So a short-haul carrier has a higher CASK even at the same efficiency, and a long-haul carrier looks cheap simply because it flies far.
The industry corrects for this with one formula: CASK is multiplied by the ratio of the carrier's stage length to a reference length, raised to the power k. The MIT Airline Data Project takes k = 0.5, uses the average passenger trip length as the length and the sample average as the reference3,26; the same exponent was used by Virgin America in correspondence with the US Securities and Exchange Commission (reference 1,000 miles), by McKinsey and by Sun Country in its 2021 prospectus2,27,28. Oliver Wyman used k = 1/3 in 20124. None of these sources explains why the square root: it is a convention. Academic work only confirms the shape of the relationship — cost per flight grows more slowly than its length29. The choice of exponent matters: for a trip twice the reference length the multiplier is 1.41 at k = 0.5 and 1.26 at k = 1/3, a difference of about 12%2,3,4,27,28,29,30,31.
CASK multiplier as a function of average passenger trip length, with a reference of 1,758 km — the median of our sample.
The chart below shows ex-fuel cost per 100 seat-kilometres for the airlines in the sample before and after adjustment. Long-haul Emirates, Air Canada and Copa look cheaper than many low-cost carriers as reported; after adjustment they move up, and short-haul European low-cost carriers move down.
Horizontal: average passenger trip length; vertical: ex-fuel cost per 100 seat-km, EUR. Circle — as reported, dot — after adjustment to the median with an exponent of 0.5.
The distance for seat-kilometres can be measured along the great circle or as actually flown, and the seats can be all seats or only those available for sale32. In 2020 Pegasus switched from great-circle to actual distance, and its 2019 seat-kilometres rose from 38.6 to 43.9 billion — by 13.9%33. With the same costs, CASK falls by 12% as a result, although nothing changed in how the airline works. Seating density also moves the denominator: by McKinsey's estimate it explains half the CASK gap on long-haul flights2.
Since 2019, under the international standard IFRS 16, a leased aircraft goes on the balance sheet and the lease payment is split into depreciation in operating expenses and interest below operating profit. The US standard keeps the whole payment for an operating lease in operating expenses34. At easyJet the transition cut aircraft lease costs from 152 to 5 million pounds, while depreciation rose from 199 to 484 million and finance costs to 39 million5. So, all else equal, a European carrier's CASK is lower than a US carrier's by the interest part of the lease.
European low-cost carriers buy fuel in advance: Ryanair hedges about 80% a year ahead19, while American had no fuel hedges at all in 20246. In years of price spikes the gap in the fuel line reflects financial decisions, not how the aircraft perform. So costs are compared ex-fuel, and fuel efficiency in physical units: grams per passenger-kilometre or passenger-kilometres per litre9,17.
Lufthansa Group reporting includes the cargo segment and third-party maintenance business35. US network carriers buy flights from regional partners at fixed rates (capacity purchase), and in their reports these flights usually sit in both costs and seat-kilometres6. Under a wet lease (ACMI), US Department of Transportation rules have the operator report the traffic and costs36. There is one rule: costs and seat-kilometres must come from the same perimeter6,36.
Converting at the current exchange rate moves costs without any change in the airline's operations; IATA warns about this explicitly1. More important: historically, low costs were more often explained by a country's cheap inputs — labour, fuel, charges — than by high efficiency37. So a CASK ranking is a ranking of cost competitiveness, not of pure efficiency. Russian carriers show this well: at the market rouble rate their costs look like a low-cost carrier's.
The choice of benchmark depends on the question. For fares and route appraisal you need the block-hour cost by aircraft type and a competitor's revenue at the same stage length. For negotiating with MRO shops and lessors, the market level of maintenance rates. For comparing airlines with one another, reports adjusted for stage length, fuel and lease accounting. For checking your own model, open series with known definitions. Quality benchmarks complement cost benchmarks, but they should not be mixed into one index: punctuality depends on airports and air traffic control as much as on the carrier16.
| Task | Which benchmarks | Why |
|---|---|---|
| Pricing and route appraisal | block-hour cost by aircraft type (Form 41 P-5.2 via FAA/ADP), competitors' reports (RASK, load factor) | gives a floor for the fare and a comparison with a competitor at the same stage length27,38 |
| Network planning and aircraft choice | Form 41 P-5.2 and T-2 by type, Swan & Adler, Wei & Hansen, ICCT | cost per seat-km depends on stage length and aircraft size at the same time17,20,21,39,40 |
| Negotiating with MRO providers and lessors | IATA MCX ($/flight hour, $/cycle), FAA (ownership costs shown separately) | shows market rate levels, but without adjusting for age and cycles1,38 |
| Investment analysis and airline comparison | reports + stage length adjustment, ex-fuel, ex-ownership or an IFRS 16 adjustment | otherwise you compare accounting standards and networks, not efficiency2,3,4,5,6,19,27,28,29,30,31,34 |
| Checking your own model (FlightCostApp) | MIT ADP / Form 41 for US types, MCX for maintenance, ICCT for fuel burn | open, independent series with known definitions1,3,17 |
| Product quality and reliability | Cirium OTP, EUROCONTROL CODA, Skytrax | complement cost metrics but do not replace them13,16,23 |
We built the rating to avoid the pitfalls as far as public reporting allows. It rests on four pillars, each answering its own question. Costs: what a seat-kilometre costs ex-fuel, adjusted to a common length by passenger trip with an exponent of 0.5 — as MIT does. Fleet productivity: how many block hours a day the aircraft fly. Commercial: load factor and operating margin — whether the airline can sell what it offers for more than it costs. Fuel: grams of jet fuel per passenger-kilometre. For each metric an airline gets a percentile within the sample — 100 for the best, 0 for the worst; the commercial pillar is the average of two metrics; the index is the average of the available pillars. Carriers with at least three pillars out of four make the rating.
Currencies are converted to euros at average ECB rates for 202589, the rouble at the average rate from Aeroflot's reports61. US carriers' utilisation is from 2025 Form 41 for their own flights, without regional partners63; fuel for some European airlines comes from CO₂ intensity per passenger-kilometre in sustainability reports, divided by 3.16 tonnes of CO₂ per tonne of jet fuel.
| # | Airline | Index | Costs ex-fuel, adjusted, per 100 seat-km | Fleet productivity block hours per aircraft | Commercial | Fuel per passenger-km | Pillars | at k = 1/3 |
|---|---|---|---|---|---|---|---|---|
| 1 | Pegasus Europe · low-cost | 90 | 932.56 EUR | —— | 81load factor 87.7 % margin 11.7 % | 9618.0 g | 3/4 | = |
| 2 | Norwegian Europe · low-cost | 78 | 753.71 EUR | 8410.9 h/day | 72load factor 85.8 % margin 10.5 % | 8122.3 g | 4/4 | ↓2 |
| 3 | Copa Lat. America · network | 78 | 644.18 EUR | 10011.9 h/day | 89load factor 87.0 % margin 22.6 % | 5825.2 g | 4/4 | ↑1 |
| 4 | Ryanair Europe · ultra-low-cost | 78 | 1002.27 EUR | 329.6 h/day | 86load factor 94.6 % margin 11.2 % | 9220.3 g | 4/4 | ↑1 |
| 5 | GOL Lat. America · low-cost | 75 | 793.46 EUR | 9011.0 h/day | 70load factor 83.8 % margin 15.2 % | 6224.6 g | 4/4 | = |
| 6 | Azul Lat. America · hybrid | 74 | 823.39 EUR | 9511.8 h/day | 64load factor 83.2 % margin 16.6 % | 5427.0 g | 4/4 | = |
| 7 | IndiGo Asia · low-cost | 73 | 962.49 EUR | 5310.1 h/day | 65load factor 84.4 % margin 9.3 % | 7722.5 g | 4/4 | = |
| 8 | Wizz Air Europe · ultra-low-cost | 71 | 892.84 EUR | 379.6 h/day | 58load factor 90.8 % margin 2.5 % | 10016.0 g | 4/4 | = |
| 9 | easyJet Europe · low-cost | 66 | 544.44 EUR | 5810.2 h/day | 65load factor 90.8 % margin 5.6 % | 8820.7 g | 4/4 | = |
| 10 | Volaris N. America · ultra-low-cost | 60 | 862.91 EUR | 218.9 h/day | 48load factor 84.3 % margin 4.4 % | 8521.0 g | 4/4 | = |
| # | Airline | Index | Costs ex-fuel, adjusted, per 100 seat-km | Fleet productivity block hours per aircraft | Commercial | Fuel per passenger-km | Pillars | at k = 1/3 |
|---|---|---|---|---|---|---|---|---|
| 11 | LATAM Lat. America · network | 52 | 614.19 EUR | —— | 75load factor 84.4 % margin 16.1 % | 1930.3 g | 3/4 | = |
| 12 | Aeroflot Group Russia · network | 50 | 574.38 EUR | 168.3 h/day | 60load factor 90.2 % margin 4.2 % | 6523.8 g | 4/4 | = |
| 13 | Frontier N. America · ultra-low-cost | 44 | 713.78 EUR | 269.1 h/day | 7load factor 78.4 % margin −4.0 % | 7322.7 g | 4/4 | = |
| 14 | Alaska + Hawaiian N. America · network | 43 | 257.18 EUR | 7910.8 h/day | 26load factor 82.9 % margin 2.1 % | 4228.0 g | 4/4 | = |
| 15 | Southwest N. America · low-cost | 38 | 366.67 EUR | 7410.3 h/day | 10load factor 77.4 % margin 1.5 % | 3129.3 g | 4/4 | ↓1 |
| 16 | Air New Zealand Oceania · network | 37 | 396.60 EUR | —— | 46load factor 83.7 % margin 7.4 % | 2729.4 g | 3/4 | ↑1 |
| 17 | Allegiant N. America · ultra-low-cost | 36 | 684.13 EUR | 07.1 h/day | 31load factor 79.3 % margin 7.8 % | 4627.9 g | 4/4 | = |
| 18 | Delta N. America · network | 33 | 09.41 EUR | 6610.2 h/day | 56load factor 83.7 % margin 9.2 % | 1232.2 g | 4/4 | ↓1 |
| 19 | Qantas Group Oceania · network | 33 | 217.44 EUR | —— | 62load factor 84.5 % margin 8.3 % | 1530.7 g | 3/4 | ↓1 |
| 20 | Sun Country N. America · leisure | 32 | 436.09 EUR | 57.1 h/day | 59load factor — margin 8.9 % | 2329.7 g | 4/4 | ↓1 |
| 21 | Air Canada N. America · network | 32 | 78.39 EUR | —— | 51load factor 84.6 % margin 4.1 % | 3828.3 g | 3/4 | ↑3 |
| 22 | Lufthansa (passenger airlines) Europe · network | 32 | 187.53 EUR | —— | 27load factor 83.2 % margin −0.4 % | 5027.0 g | 3/4 | = |
| 23 | JetBlue N. America · hybrid | 31 | 326.96 EUR | 429.9 h/day | 14load factor 82.4 % margin −4.1 % | 3529.0 g | 4/4 | = |
| 24 | United N. America · network | 31 | 118.38 EUR | 6610.2 h/day | 38load factor 82.2 % margin 8.0 % | 832.3 g | 4/4 | = |
| 25 | American N. America · network | 25 | 147.88 EUR | 479.9 h/day | 35load factor 83.6 % margin 2.7 % | 433.7 g | 4/4 | = |
| 26 | airBaltic Europe · hybrid | 23 | 464.80 EUR | 107.7 h/day | 12load factor 80.2 % margin −3.4 % | —— | 3/4 | = |
| 27 | Finnair Europe · network | 11 | 296.96 EUR | —— | 5load factor 76.9 % margin −1.6 % | 034.7 g | 3/4 | = |
Scores are percentiles within the sample (100 is best); metrics are the underlying figures: ex-fuel costs in EUR per 100 seat-km, adjusted to a passenger trip of 1,758 km. “At k = 1/3” shows how far a place moves if costs are adjusted with an exponent of 1/3 instead of 0.5. Not rated: Aegean — data for only 2 pillars out of four; Emirates — data for only 2 pillars out of four; Spirit Airlines (Spirit Aviation Holdings) — two bankruptcies in 2025, the year is split into two reporting periods; S7 Airlines — data for only 2 pillars out of four; WestJet — no public financial reporting (secondary sources only).
| Airline | Period | Ex-fuel costs, adjusted, per 100 seat-km | Passenger trip, km | Utilisation, h/day | Load factor | Margin | Fuel, g / passenger-km | Sources |
|---|---|---|---|---|---|---|---|---|
| Pegasus | FY2025: calendar year | 2.56 EUR | 1,587 | — | 87.7 % | 11.7 % | 18.0 | 41,42 |
| Norwegian | FY2025: calendar year | 3.71 EUR | 1,398 | 10.9 | 85.8 % | 10.5 % | 22.3 | 43 |
| Copa | FY2025: calendar year | 4.18 EUR | 3,057 | 11.9 | 87.0 % | 22.6 % | 25.2 | 44 |
| Ryanair | FY2025 (April 2024 – March 2025) | 2.27 EUR | 1,262 | 9.6 | 94.6 % | 11.2 % | 20.3 | 45,46,47 |
| GOL | FY2025: calendar year | 3.46 EUR | 1,197 | 11.0 | 83.8 % | 15.2 % | 24.6 | 48,49 |
| Azul | FY2025: calendar year | 3.39 EUR | 1,327 | 11.8 | 83.2 % | 16.6 % | 27.0 | 50 |
| IndiGo | FY2026: 1 Apr 2025 – 31 Mar 2026 | 2.49 EUR | 1,179 | 10.1 | 84.4 % | 9.3 % | 22.5 | 51,52,53,54 |
| Wizz Air | F26: 1 Apr 2025 – 31 Mar 2026 | 2.84 EUR | 1,719 | 9.6 | 90.8 % | 2.5 % | 16.0 | 55,56 |
| easyJet | FY25: 1 Oct 2024 – 30 Sep 2025 | 4.44 EUR | 1,306 | 10.2 | 90.8 % | 5.6 % | 20.7 | 57,58 |
| Volaris | FY2025 | 2.91 EUR | 1,581 | 8.9 | 84.3 % | 4.4 % | 21.0 | 59 |
| LATAM | FY2025: calendar year | 4.19 EUR | 1,649 | — | 84.4 % | 16.1 % | 30.3 | 60 |
| Aeroflot Group | 2025 (calendar year) | 4.38 EUR | 2,784 | 8.3 | 90.2 % | 4.2 % | 23.8 | 61 |
| Frontier | FY2025 | 3.78 EUR | 1,512 | 9.1 | 78.4 % | −4.0 % | 22.7 | 62,63 |
| Alaska + Hawaiian | FY2025 | 7.18 EUR | 2,117 | 10.8 | 82.9 % | 2.1 % | 28.0 | 63,64 |
| Southwest | FY2025 | 6.67 EUR | 1,673 | 10.3 | 77.4 % | 1.5 % | 29.3 | 63,65 |
| Air New Zealand | FY26: 1 Jul 2025 – 30 Jun 2026 | 6.60 EUR | 2,145 | — | 83.7 % | 7.4 % | 29.4 | 66,67 |
| Allegiant | FY2025 | 4.13 EUR | 1,456 | 7.1 | 79.3 % | 7.8 % | 27.9 | 63,68 |
| Delta | FY2025 (1 Jan – 31 Dec 2025) | 9.41 EUR | 2,008 | 10.2 | 83.7 % | 9.2 % | 32.2 | 63,69,70 |
| Qantas Group | FY26: 1 Jul 2025 – 30 Jun 2026 | 7.44 EUR | 2,384 | — | 84.5 % | 8.3 % | 30.7 | 71 |
| Sun Country | FY2025 | 6.09 EUR | 2,052 | 7.1 | — | 8.9 % | 29.7 | 63,72 |
| Air Canada | FY2025 | 8.39 EUR | 3,163 | — | 84.6 % | 4.1 % | 28.3 | 73 |
| Lufthansa (passenger airlines) | FY2025: calendar year | 7.53 EUR | 2,087 | — | 83.2 % | −0.4 % | 27.0 | 74,75,76,77 |
| JetBlue | FY2025 | 6.96 EUR | 2,190 | 9.9 | 82.4 % | −4.1 % | 29.0 | 63,78 |
| United | FY2025 | 8.38 EUR | 2,414 | 10.2 | 82.2 % | 8.0 % | 32.3 | 63,79 |
| American | FY2025 | 7.88 EUR | 1,798 | 9.9 | 83.6 % | 2.7 % | 33.7 | 63,80,81 |
| airBaltic | FY2025: calendar year | 4.80 EUR | 1,405 | 7.7 | 80.2 % | −3.4 % | — | 82 |
| Finnair | FY2025: calendar year | 6.96 EUR | 2,540 | — | 76.9 % | −1.6 % | 34.7 | 83,84 |
| Aegean | FY2025: calendar year | 4.44 EUR | 1,055 | — | 82.2 % | 11.7 % | — | 85 |
| Emirates | FY2025-26: 1 Apr 2025 – 31 Mar 2026 | 8.93 EUR | 5,246 | — | 78.4 % | 17.8 % | — | 86,87 |
| S7 Airlines | 2025 | — | 2,642 | — | 94.2 % | 11.3 % | 23.3 | 88 |
At the top are airlines strong on several pillars at once. The leader, Pegasus, has three pillars out of four: its utilisation could not be read reliably from its report, so its place should be seen as provisional. Pegasus and Ryanair combine some of the lowest costs and good fuel efficiency with high load factors; Copa, with its hub in Panama, has the highest margin in the sample (22.6 %) and high utilisation; Norwegian has even results across all pillars. Next come the Latin American and Indian low-cost carriers — GOL, Azul, IndiGo — with low costs and high utilisation. The business model does not guarantee a place: Frontier and Allegiant are ultra-low-cost carriers, but in 2025 the first had a negative margin and a low load factor, and the second one of the lowest utilisation figures in the sample.
At the bottom are the large network carriers of the US and Europe. Their cost per seat-kilometre is one and a half to two times higher, with average margins and utilisation; that is the price of a hub network with premium cabins and short connecting flights. An important caveat: US network carriers' costs include their regional partners' flights — at American, for example, $5.1 billion80 — while we take their utilisation from their own flights only. Finnair comes last: high costs over a long passenger trip, a low load factor, a loss and the highest fuel burn in the sample.
Aeroflot is no. 12. Its ex-fuel costs at the market rouble rate are 3.48 EUR / 100 seat-km, like a low-cost carrier's, but that is exactly the currency pitfall: cheap inputs, not efficiency. After adjusting for length (the group's passenger trip is 2,784 km) costs rise to the middle of the sample. The load factor is among the highest, 90.2 %; utilisation is 8.3 hours a day across the whole fleet on the balance sheet, below most; the margin is 4.2 %. S7 is not in the rating: its accounts do not separate fuel from cost of sales, so the cost pillar cannot be calculated88.
Four researchers kept logs — 1,061 steps (trace/b1–b7): the review of benchmarks and pitfalls, data for airlines outside and inside North America, and filling in fuel and utilisation; extracts from Form 41, Aeroflot's Databook and exchange rates are in separate logs. Separate agents opened every source and checked the quotes (verify/): review — 112 found verbatim; airlines outside North America — 227; North America — 196; not a single number differed from its source.
Corrected after checking: after Pegasus changed its seat-kilometre definition, CASK falls by 12%, not 13.9%; MIT adjusts costs to the passenger trip — and so does our rating. Estimates rather than reported figures are marked in the source table: the average fleet of Copa, Azul and airBaltic; Norwegian's and Aegean's costs as revenue minus operating profit; IndiGo's, GOL's and Air New Zealand's fuel recalculated from energy or CO₂. Spirit is excluded: two bankruptcies and two reporting periods in 2025. Ryanair is for the financial year to March 2025: the following year's report has no seat-kilometres.