Is EROI real? The ratio is real, but there is no single number
By OilPriceAPI · Published · Thirteen journal papers published from 2009 to 2024, read Nov 13, 2026. The newest observation in them is for 2020.
EROI, energy return on investment, is the energy a process delivers divided by the energy spent to get it. It is a real physical ratio and it can be estimated for one oil field or for a whole industry. What does not exist is one current number for oil. Published estimates for the world's fossil fuels run from about 30:1 where the fuel leaves the ground (Brockway, Owen, Brand-Correa and Hardt (2019), for 1995 to 2011) to about 3.5:1 after the engine or boiler has used it (Aramendia, Brockway, Taylor, Norman, Heun and Marshall (2024), for 2020).
Those two figures do not contradict each other. They count energy at different points in the same chain. This article sets out what each published estimate measures, why estimates for the same fuel differ, what the evidence says about the trend, and what EROI does and does not say about prices or supply. It gives the case that EROI is falling and matters, and the case that it is not a usable gauge, each in its strongest form.
What EROI measures
Gagnon, Hall and Brinker (2009) define it this way: “EROI is the ratio of energy that is produced by a process (in this case oil and gas extraction) to the energy that is consumed in carrying out that process.” Hall, Balogh and Murphy (2009) give the reading of a ratio of 30:1: it “implies that a particular process yields 30 Joules on an investment of 1 Joule”. The ratio has no units. The same idea is also written EROEI, energy return on energy invested.
The share of gross energy left over after the energy spent is 1 minus 1/EROI. The table applies that formula to six ratios that appear in the papers below. It is arithmetic on the definition and is not a figure from any paper.
| EROI | Energy spent, % of gross | Energy left, % of gross |
|---|---|---|
| 35:1 | 2.9 | 97.1 |
| 18:1 | 5.6 | 94.4 |
| 8.5:1 | 11.8 | 88.2 |
| 6:1 | 16.7 | 83.3 |
| 3.5:1 | 28.6 | 71.4 |
| 2:1 | 50.0 | 50.0 |
Energy left = (1 - 1/EROI) x 100. Our arithmetic. The ratios come from different papers and different boundaries, listed in the next table, so the rows are not a time series.
The relationship is not a straight line. Going from 35:1 to 18:1 changes the energy left from 97.1% to 94.4%, a difference of 2.7 points. Going from 6:1 to 3.5:1 changes it from 83.3% to 71.4%, a difference of 11.9 points. A fall in the ratio matters little while the ratio is high and a great deal once it is low. The title of figure 6 in Brockway, Owen, Brand-Correa and Hardt (2019) calls this shape the “net energy cliff”.
Why estimates for the same fuel differ so much
Four choices set the number, and papers make them differently.
- Where the output is counted. Hall, Lambert and Balogh (2014) separate a standard EROI, taken where the fuel “leaves the extraction or production facility (well-head, mine mouth, farm gate, etc.)”, from a point-of-use EROI that also counts refining and transport, and an extended EROI that also counts the energy needed to use the fuel. In their words: “As the boundaries of the analysis are expanded, the energy cost of getting it to that point increases, resulting in a reduced EROI”. Brockway, Owen, Brand-Correa and Hardt (2019) use the terms primary stage and final stage for the first two, and Aramendia, Brockway, Taylor, Norman, Heun and Marshall (2024) add a useful stage, after the end-use device.
- Which inputs are counted. The standard EROI counts energy used on site and the energy used to make the materials bought. It leaves out the energy behind labour and financial services (Hall, Lambert and Balogh (2014)). Some studies also leave out the energy used to build equipment: Brandt, Sun, Bharadwaj, Livingston, Tan and Gordon (2015) say their model “does not currently include energy invested in building oilfield capital equipment”, and Aramendia, Brockway, Taylor, Norman, Heun and Marshall (2024) call their values “an upper bound excluding capital investments”.
- How the inputs are obtained. Few countries record the energy their oil industry uses. Guilford, Hall, O'Connor and Cleveland (2011) use US census and EIA data. Gagnon, Hall and Brinker (2009) had only company spending in dollars for the world, and converted dollars to energy. Court and Fizaine (2017) go back to 1860 by calculating EROI from fuel prices.
- What is added together. Most series are for oil and gas combined, because the two come from the same wells. A highlight of Hall, Lambert and Balogh (2014) reads: “Lower EROI for oil may be masked by natural gas extracted/used in oil production.” Other series are for all fossil fuels, or for one field.
The researchers who built the field say this plainly. Murphy, Hall, Dale and Cleveland (2011) wrote a protocol because “the literature to date is in fact confusing, divergent, and often acrimonious”, and their instruction to authors is: “The important thing is to make what you include very clear.” Raugei (2019) adds that EROI “must not be used to compare carriers that cannot be put to similar end-use”, and that combining carriers into an average or minimum EROI “can be problematic”.
Even one study gets restated differently. The abstract of Gagnon, Hall and Brinker (2009) gives 26:1 in 1992, 35:1 in 1999 and 18:1 in 2006. Hall, Lambert and Balogh (2014), with one author in common, describe the same result in one passage as a decline “from 30:1 in 1995 to about 18:1 in 2006” and in another as “approximately 23:1 in 1992, 33:1 in 1999 and 18:1 in 2005”. An EROI figure quoted second hand should be checked against the paper it came from.
The published estimates, with boundaries
| Study | What was measured | Boundary and method | Published estimate |
|---|---|---|---|
| Gagnon, Hall and Brinker (2009) | Oil and gas together, publicly traded companies worldwide, 1992 to 2006 | At the wellhead. Energy inputs estimated from company spending in dollars, converted to energy | 26:1 in 1992, 35:1 in 1999, 18:1 in 2006 |
| Guilford, Hall, O'Connor and Cleveland (2011) | US oil and gas production, 1919 to 2007 | Production by the oil and gas industry, from US Census of Mineral Industries and EIA data | About 20:1 from 1919 to 1972, about 8:1 in 1982, about 17:1 from 1986 to 2002, about 11:1 in the mid to late 2000s |
| Court and Fizaine (2017) | World oil from 1860 and world gas from 1890, each to 2012 | Primary production. Calculated from fuel prices, not from measured energy inputs | Peaks in the 1930s to 1940s, around 50:1 for oil and around 150:1 for gas, lower since |
| Brockway, Owen, Brand-Correa and Hardt (2019) | All fossil fuels, world, 1995 to 2011 | Primary stage (fuel as extracted) and final stage (electricity and petrol as delivered) | About 30:1 at the primary stage; around 6:1 and declining at the final stage |
| Aramendia, Brockway, Taylor, Norman, Heun and Marshall (2024) | All fossil fuels, world, 1971 to 2020 | Final stage and useful stage (after the engine, boiler or lamp). Energy used to build industry capital is not counted | Final stage about 9.6:1 in 1971 and 8.5:1 in 2020; useful stage 2.9:1 in 1971 and 3.5:1 in 2020 |
| Delannoy, Longaretti, Murphy and Prados (2021) | All oil liquids, world, modelled 1950 to 2050 | Standard EROI at the wellhead, direct and indirect inputs. A model built on literature values, not a measurement | Weighted average falls from a maximum of 44.4 to a modelled plateau of 6.7 |
| Brandt (2011) | 306 California oil fields, 1955 to 2005 | Life cycle: energy delivered to final consumers per unit of primary energy spent on extraction, transport and refining | About 6.5:1 in 1955, about 3.5:1 in 2005 |
| Brandt, Sun, Bharadwaj, Livingston, Tan and Gordon (2015) | 40 oilfields worldwide, field by field | Crude oil produced per unit of all fuels consumed in drilling, producing, processing and transport. Energy to build equipment is not counted | From about 2:1 to about 100:1, depending on the field |
| Hall, Lambert and Balogh (2014) | Review of published estimates: world oil and gas, tar sands, oil shale | Mean of values published in earlier studies, collected by the review | World oil and gas about 20:1 (36 values from 4 publications); tar sands 4:1 (4 values); oil shale 7:1 (15 values) |
Each figure is the paper's own, at the paper's own boundary and for the years shown. Full references are under Sources. Read Nov 13, 2026. Figures from different rows should not be subtracted or averaged.
The case that EROI is falling and matters
We set this out as the case a reader who expects limits could make. It is not a quotation, except where marked.
- EROI is physical. A price can be moved by currencies, taxes or a war. The energy needed to lift, process and deliver a barrel cannot.
- Every oil or gas time series in the table ends below its high. World oil and gas at the wellhead went from 35:1 in 1999 to 18:1 in 2006. US production went from about 20:1 before 1972 to about 11:1 in the mid to late 2000s. The price-based world series peaks in the 1930s to 1940s. California went from about 6.5:1 to about 3.5:1 over 50 years. The final-stage series for all fossil fuels is “around 6:1 and declining”.
- The oil that replaces old fields returns less. The review by Hall, Lambert and Balogh (2014) gives a mean of 4:1 for tar sands against about 20:1 for world oil and gas, and concludes: “The decline in EROI among major fossil fuels suggests that in the race between technological advances and depletion, depletion is winning.”
- Low ratios are where the loss is large. At 6:1 the energy left is 83.3% of gross and at 3.5:1 it is 71.4%. Brockway, Owen, Brand-Correa and Hardt (2019) warn that final-stage ratios “could decline precipitously in the near future”.
- A model built on these values projects the trend forward. Delannoy, Longaretti, Murphy and Prados (2021) estimate that producing oil liquids took energy equal to 15.5% of their gross energy output at the time of writing, and project about half by 2050. That is a model result, discussed below.
The case that EROI is not a usable gauge
This is the case a reader who expects abundance could make, also in our words.
- There is no agreed number to watch. For fossil fuels the published figure is about 30:1, about 8.5:1, around 6:1 or about 3.5:1 depending on the stage and the method. A gauge that reads four ways cannot signal a threshold.
- The longest physical series shows a slow decline. In Aramendia, Brockway, Taylor, Norman, Heun and Marshall (2024), built on International Energy Agency energy balances, the final-stage EROI of fossil fuels went from about 9.6:1 in 1971 to 8.5:1 in 2020. That is a fall of 11.5% in 49 years.
- At the point where energy does work, the ratio rose. The same paper finds the useful-stage EROI of fossil fuels went from 2.9:1 to 3.5:1, and of oil products from 1.96:1 to 2.04:1, between 1971 and 2020. The authors' explanation: “These trends are due to the effect of increasing final-to-useful efficiencies offsetting the decrease in final-stage EROIs.” They add that the findings “contradict the view according to which fossil fuels have high, although rapidly decreasing, net energy returns”.
- Two of the series with the largest swings are built from money, not from measured energy. The world wellhead series converts company spending to energy and the 1860 to 2012 series is calculated from prices. A series built from prices or spending moves when prices and spending move.
- Fields differ far more than the average has moved. Across 40 oilfields Brandt, Sun, Bharadwaj, Livingston, Tan and Gordon (2015) find ratios “from ≈2 to ≈100”, and say the “lowest energy returns are found to come from thermally-enhanced oil recovery technologies”. What is being produced, and how, explains more than the calendar year.
- The decline has reversed before. The world wellhead series rose from 26:1 in 1992 to 35:1 in 1999 before it fell, and US production recovered from about 8:1 in 1982 to about 17:1 from 1986 to 2002.
What the evidence says about the trend
On direction the papers agree for the wellhead and the final stage: every series we read is lower at its end than at its start or its earlier high. The useful stage is the exception, and it rose.
On size they do not agree. The world wellhead series falls 48.6% in seven years (35:1 to 18:1, 1999 to 2006). The final-stage series built on energy balances falls 11.5% in 49 years (about 9.6:1 to 8.5:1, 1971 to 2020). These measure different things, so the two rates cannot be ranked as right and wrong. They can only be read with their boundaries.
The ratio also moves with drilling. Guilford, Hall, O'Connor and Cleveland (2011) report “a gradual secular decrease in EROI and an inverse relation to drilling effort”, and write that “the lower the intensity of drilling, the higher the EROI compared to the secular trend”. Gagnon, Hall and Brinker (2009) find the same for the world: EROI declines “over time and with increased annual drilling levels”. Part of each fall is depletion and part is the industry drilling harder when prices are high.
Two papers from overlapping authors give different final-stage levels: around 6:1 for 1995 to 2011 (Brockway, Owen, Brand-Correa and Hardt (2019)) and about 8.5:1 for 2020 (Aramendia, Brockway, Taylor, Norman, Heun and Marshall (2024)). The later paper uses a different accounting framework and does not count capital investment. We could read only the abstract of the earlier paper, so we do not attribute the gap to one cause.
There are three gaps. No series we found goes past 2020. No official body maintains one: Hall, Lambert and Balogh (2014) present their values “with considerable humility because there are no government-sponsored programs or much financial support to derive such numbers”. And the review by Murphy and Hall (2010) closes, in the words of its abstract, “by listing numerous areas of improvement that are needed within EROI research”.
What EROI does and does not say about prices and supply
- It is not an independent predictor of price where it was computed from price. Court and Fizaine (2017) estimate each year's EROI as the energy sector's monetary return on investment divided by the fuel price times the energy intensity of spending (equation 4 of their working paper). Other things equal, a higher price gives a lower EROI by construction. The working paper says readers “shall retain orders of magnitude and trends and absolutely not precise estimated values for given years”.
- It is partly a result of price. When prices rise the industry drills more, and both long series above show EROI falling as drilling rises. A low EROI in a boom year is not by itself evidence of scarcity.
- It does not date a peak. Delannoy, Longaretti, Murphy and Prados (2021) take a forecast of gross oil liquids production from GlobalShift (2020), which peaks in 2034 at 551 PJ a day, and subtract the energy cost. Their modelled net energy peaks in 2024 at 415 PJ a day, with a standard deviation of 6.6 years across scenarios. The authors write: “This first result should not be interpreted as the announcement of a coming peak”. It is a projection that later data can test. It is not an observation.
- It does not set a floor that current estimates are near. Hall, Balogh and Murphy (2009) calculate a minimum EROI for oil and for corn ethanol “as about 3:1 at the mine-mouth/farm-gate”. The measured wellhead figures in the table for world oil and gas are 18:1 and higher, the US low is about 8:1 in 1982, and the modelled plateau for oil liquids is 6.7. The figures near 3:1 are for tar sands (a mean of 4:1) and for boundaries further down the chain, which cannot be compared with a mine-mouth threshold.
- It says nothing about 2026 prices. The EIA's Short-Term Energy Outlook released Nov 10, 2026 attributes October's rise in crude prices to “falling global inventories, attacks on oil tankers in the Persian Gulf, constrained shipping capacity, and higher transportation costs”. The newest EROI observation is for 2020, so no EROI series can be tested against this year's prices.
- It does not measure how much oil is in the ground, and it is not investment advice.
What would change the verdict
The scarcity reading would gain support from a series that states its boundary, is updated every year by the same method, and keeps falling at the final stage through the 2020s, with the useful stage starting to fall as well. It would gain more if the fall continued in years without a supply disruption, when drilling effort is not unusually high.
The abundance reading would gain support if the same kind of series stayed flat or rose after 2020, or if field-level studies showed new supply returning as much energy as the fields it replaces. Neither series exists yet in a form we could read. The modelled 2024 net-energy peak for oil liquids is the nearest testable claim, and it needs measured data for 2024 and later to test it.
Verdict
EROI is real as a concept and as a measurement. It is not a single number, and there is no current or official one for oil. The published evidence supports a decline for oil and gas at the wellhead and at the final stage. It does not support a precise rate: the estimates range from 11.5% in 49 years to 48.6% in seven, on different boundaries. At the useful stage the longest series rose slightly to 2020.
The limits of this verdict: for three of the thirteen papers only the abstract or summary was open to us, the newest data are six years old, and two of the series are derived from prices or spending. Anyone who quotes one EROI figure for oil without its boundary, method and year is quoting less than the literature supports.
Related: The Jevons check: world energy in 2025 · Oil priced in gold · Has the Permian peaked?
Sources
- Gagnon, Hall and Brinker (2009). “A Preliminary Investigation of Energy Return on Energy Investment for Global Oil and Gas Production.” Energies 2, 490-503. doi:10.3390/en20300490. What we read: open-access article, abstract and introduction.
- Hall, Balogh and Murphy (2009). “What is the Minimum EROI that a Sustainable Society Must Have?.” Energies 2, 25-47. doi:10.3390/en20100025. What we read: open-access article, abstract and introduction.
- Murphy and Hall (2010). “Year in review: EROI or energy return on (energy) invested.” Annals of the New York Academy of Sciences. doi:10.1111/j.1749-6632.2009.05282.x. What we read: publisher page, abstract only (the article is paywalled).
- Guilford, Hall, O'Connor and Cleveland (2011). “A New Long Term Assessment of Energy Return on Investment (EROI) for U.S. Oil and Gas Discovery and Production.” Sustainability 3(10), 1866. doi:10.3390/su3101866. What we read: open-access article, abstract.
- Murphy, Hall, Dale and Cleveland (2011). “Order from Chaos: A Preliminary Protocol for Determining the EROI of Fuels.” Sustainability 3(10), 1888. doi:10.3390/su3101888. What we read: open-access article, abstract, introduction and section 2.
- Brandt (2011). “Oil Depletion and the Energy Efficiency of Oil Production: The Case of California.” Sustainability 3(10), 1833. doi:10.3390/su3101833. What we read: open-access article, abstract.
- Hall, Lambert and Balogh (2014). “EROI of different fuels and the implications for society.” Energy Policy 64, 141-152. doi:10.1016/j.enpol.2013.05.049. What we read: open-access article, full text (PDF of the published article, sections 1.2, 2 and 4.1).
- Brandt, Sun, Bharadwaj, Livingston, Tan and Gordon (2015). “Energy Return on Investment (EROI) for Forty Global Oilfields Using a Detailed Engineering-Based Model of Oil Production.” PLOS ONE 10, e0144141. doi:10.1371/journal.pone.0144141. What we read: open-access article, abstract.
- Court and Fizaine (2017). “Long-Term Estimates of the Energy-Return-on-Investment (EROI) of Coal, Oil, and Gas Global Productions.” Ecological Economics 138, 145-159. doi:10.1016/j.ecolecon.2017.03.015 (open-access copy). What we read: abstract of the published article (HAL record hal-01549813) and the authors' open-access working paper (HAL hal-03186990, IFP School Cahiers de l'Economie no. 113), full text.
- Raugei (2019). “Net energy analysis must not compare apples and oranges.” Nature Energy 4, 86-88. doi:10.1038/s41560-019-0327-0. What we read: publisher page, summary only (the article is paywalled).
- Brockway, Owen, Brand-Correa and Hardt (2019). “Estimation of global final-stage energy-return-on-investment for fossil fuels with comparison to renewable energy sources.” Nature Energy 4, 612-621. doi:10.1038/s41560-019-0425-z. What we read: publisher page, abstract only (the article is paywalled).
- Delannoy, Longaretti, Murphy and Prados (2021). “Peak oil and the low-carbon energy transition: A net-energy perspective.” Applied Energy 304. doi:10.1016/j.apenergy.2021.117843 (open-access copy). What we read: authors' open-access copy on HAL (hal-03360253, CC BY 4.0), full text, sections 3.3 and 4.1.
- Aramendia, Brockway, Taylor, Norman, Heun and Marshall (2024). “Estimation of useful-stage energy returns on investment for fossil fuels and implications for renewable energy systems.” Nature Energy 9, 803-816. doi:10.1038/s41560-024-01518-6. What we read: open-access article, full text.
- US Energy Information Administration, Short-Term Energy Outlook, released Nov 10, 2026, forecast overview. Read Nov 13, 2026.