Refinery Crack Spreads, Explained

1. What a crack spread actually is

A refinery does one job: it buys crude oil, takes it apart, and sells the pieces for more than it paid. Heat, pressure, and catalysts "crack" the long hydrocarbon chains in crude into shorter ones, turning a heavy, low-value barrel into gasoline, diesel, jet fuel, and heating oil. The crack spread is the market's real-time price tag on that transformation: the value of the products coming out, minus the cost of the crude going in.

That is all it is. It is not a company profit figure, not a forecast, and not a physical quantity. It is a margin proxy, priced every trading day by the futures market, that tells you whether refining is currently a good business or a painful one. When the spread is wide, refiners are capturing a large processing margin and will run their plants as hard as they can. When it is narrow, they are barely covering costs, and capacity starts shutting in.

This distinction matters more than most oil commentary admits. The price of crude tells you about the cost of the raw material. The crack spread tells you about the margin, which is the part that actually shows up as refiner profit and, eventually, as the gap between crude prices and what you pay at the pump. In 2026 this distinction has mattered more than at almost any point in the last two decades: crude prices swung on geopolitics, but the spread told a completely different story about where the real shortage sat.

The industry's standard shorthand is the 3-2-1 crack spread: buy three barrels of crude, sell two barrels of gasoline and one barrel of distillate. That ratio approximates the product yield of a typical US refinery, so the 3-2-1 has become the benchmark quote for American refining margins, the number traders, analysts, and refiner earnings calls all reference.

2. The arithmetic: 3 barrels in, 3 barrels out

The formula looks fussy the first time because the units do not match. Crude oil futures are quoted in dollars per barrel. Gasoline and heating oil futures are quoted in dollars per gallon. Since one barrel holds 42 gallons, every product price gets multiplied by 42 before the comparison:

3-2-1 crack = ((2 × gasoline $/gal × 42) + (1 × distillate $/gal × 42) − (3 × crude $/bbl)) ÷ 3

The heating oil leg stands in for diesel and the broader distillate pool (diesel, jet fuel, heating oil all come from the same middle of the distillation column and trade closely together). The final division by three converts the total back into dollars per barrel of crude, so the answer reads as a per-barrel margin.

Here is a worked example with hypothetical teaching figures, chosen for clean arithmetic, not descriptions of any real market day:

A $22.40 result means the market is pricing a gross margin of $22.40 on each barrel of crude run through the 3-2-1 recipe. "Gross" is doing heavy lifting in that sentence: it is revenue minus feedstock only. Labor, utilities, catalysts, maintenance, compliance costs, and debt service all still come out of it. A teaching text on refinery economics puts those variable operating costs in the ballpark of $20 per barrel under typical conditions, which is why a $22 crack is comfortable while a $10 crack is a survival question. Real operating costs vary widely by plant, region, and energy prices, so treat any single figure as illustrative, not as a refinery's actual cost.

3. The spread family: 2-1-1, 5-3-2, and single-product cracks

The 3-2-1 is the headline quote, but it is one member of a family, and each sibling answers a different question:

• 2-1-1 crack. One barrel of gasoline plus one barrel of distillate, minus two barrels of crude, divided by two. A balanced gasoline-to-distillate proxy for refiners whose output mix is less gasoline-heavy than the 3-2-1 assumes.
• 5-3-2 crack. Three barrels of gasoline and two of distillate from five barrels of crude, a gasoline-heavy recipe. This is the standard quote for West Coast margins, where the CARB reformulated gasoline specification dominates the product slate.
• 1-1 (single-product) cracks. Just gasoline minus crude, or just diesel minus crude. Crude when you want to isolate which product is driving the margin story, and 2026 has been a year when that distinction carried the whole analysis.

The benchmark legs also change by region. The futures-market 3-2-1 uses WTI crude, RBOB gasoline, and heating oil futures on NYMEX, and exchanges list the whole basket as a single tradeable spread contract. The US Energy Information Administration's teaching example, by contrast, builds a Gulf Coast 3-2-1 from Louisiana Light Sweet crude plus Gulf Coast conventional gasoline and ultra-low sulfur diesel spot prices, because that matches what an actual Gulf Coast refinery buys and sells. Same idea, different basis. Which version you are looking at matters whenever regional prices diverge, which is the subject of section 7.

4. How to read the number: from healthy to acute stress

A crack spread quote means nothing without a sense of scale. Industry references, including a March 2026 MUFG market note built on Bloomberg data, read the WTI 3-2-1 roughly like this:

Against that yardstick, 2026 has lived almost entirely in the stress zones. The WTI 3-2-1 climbed from under $20 per barrel to over $54 early in the year, jumped about 49 percent after the March 1 Iran strikes, and sat above $40, the acute-stress line. It reached roughly $55.78 in March, a four-year high, then kept climbing to a record closing level of $69.66 per barrel on July 16, traded between about $56 and $65 through the summer, and sat near $73 in late September. For perspective, the all-time record for the 3-2-1, set on May 13, 2022, is $75.89. Late-2026 readings have been within striking distance of that mark for months.

The direction of the move tells you where the problem is. In a normal crude supply shock, oil prices jump and the crack spread compresses, because refiners cannot pass the higher feedstock cost through to products immediately. When crude spikes but the crack spread widens anyway, the shock is coming from the product side: refineries, inventories, or product trade flows, not the wellhead. That is the signature of 2026, and it is why the spread has been the single most informative number in oil markets this year.

5. Why the spread spikes

Crack spreads widen when product prices outrun crude prices. That happens through a short list of recurring mechanisms, and learning to tell them apart is most of the practical skill in reading the spread:

• Turnarounds. Refineries shut units for planned maintenance, usually in spring and fall, the shoulder seasons between winter heating and summer driving demand. A heavy turnaround season takes several percent of capacity offline at once. In 2026 the effect was amplified because operators deferred maintenance to keep running through strong margins: industry trackers counted about $1.1 billion in US refinery maintenance scheduled for the first quarter, with several plants pushing work back rather than taking units down while cracks were this wide. Deferred maintenance is not canceled maintenance. Every pushed turnaround is a future outage accumulating.
• Hurricanes and unplanned outages. Gulf Coast refineries sit in hurricane country, and a single storm can idle millions of barrels per day of capacity. The spread spikes first, then fades as plants restart. Short, sharp, and the classic reason the 3-2-1 has a long history of violent but brief excursions above $30.
• Driving season. US gasoline demand peaks in summer, and gasoline cracks reliably firm into Memorial Day. This is the seasonal component of the spread, predictable enough that analysts deseasonalize the series before looking for real dislocations.
• Product inventory draws. When distillate or gasoline stocks fall well below seasonal norms, every marginal barrel of demand bids up products faster than crude. US distillate stocks hit a 30-year seasonal low of 104.2 million barrels in late August 2026, which is a textbook setup for sustained wide cracks.
• Structural capacity loss. Refineries that close do not reopen. Europe has been shedding capacity for years, and each closure removes the spare cushion that used to absorb the shocks above. With US refineries running at 97 to 98 percent of capacity in 2026 against a typical rate near 90 percent, there is almost nothing left in reserve if another plant goes down.

Note what is missing from the list: crude oil prices. A pure crude rally, with products lagging, narrows the crack. That is why the spread is such a useful diagnostic. It separates a crude problem from a refining problem, and the policy and investment responses to those two problems are completely different.

6. 2026: the year distillate broke the model

The 2026 crack spread story is really two stories, and the blended 3-2-1 hides the more important one. Gasoline margins surged first, with the gasoline crack topping about $59 per barrel in July before moderating to roughly $38.90 by September. Distillate margins went the other way: the diesel crack cleared $100 per barrel for the first time in mid-August, pushed above $106 on September 1, a record, and touched an intraday record of $112.76 before settling near $108. For context, the diesel crack has historically averaged closer to $19 per barrel going back to 2006, and normal years produce $20 to $30.

The September 1 arithmetic is worth seeing in full, because it shows what a product-side shortage looks like. Gulf Coast ultra-low sulfur diesel settled at $4.734 a gallon that day, which is $198.83 per barrel. WTI crude fetched $91.48. The gap works out to nearly $107 a barrel of processing margin, or about $2.52 of margin in every single gallon of diesel. In an ordinary year that figure runs 50 to 70 cents a gallon.

Meanwhile the blended 3-2-1 sat at $64.34 on that same day, about 15 percent below its 2022 record, at the exact moment the diesel margin was setting new records. That divergence is the proof of the diagnosis: there was no general refining shortage and no general energy shortage in September 2026. There was a shortage in precisely one cut of the barrel, middle distillate. Diesel, heating oil, marine gasoil, and jet fuel all come out of the same section of the distillation column, compete for the same equipment, and move together.

The drivers stacked up through the year. The Strait of Hormuz has been effectively closed since late February, pushing Brent from $61 at the start of the year to over $118 by the end of the first quarter. Ukrainian drone strikes disrupted an estimated 20 to 40 percent of Russia's refining capacity, and Moscow banned diesel exports through September 30. China and India both limited their own diesel exports. US distillate inventories sank to that 30-year seasonal low even as refineries ran at 97 to 98 percent. Foreign buyers paid up for American diesel, so cargoes left the Gulf Coast for export rather than refilling domestic tanks. The EIA, which had earlier forecast a $1.30 per gallon diesel spread, raised it to $1.57, and projected the diesel crack would stay above $2 a gallon through November, with distillate stocks forecast to fall below 100 million barrels.

Refiner earnings told the same story in dollars. Marathon Petroleum's refining and marketing margin rose to $36.33 per barrel in the second quarter from $17.58 a year earlier, with net income going from $1.2 billion to $5.1 billion. Valero's refining margin went from $11.78 to $24.42, with refining operating income up from $1.3 billion to $4.5 billion. Phillips 66's margin went from $11.25 to $24.08, and its net income from $0.9 billion to $3.8 billion. Both Marathon and Phillips 66 ran their refineries at 94 to 96 percent of capacity. Management teams attributed the improvement directly to higher market crack spreads, which is about as clean a confirmation as a margin proxy ever gets.

7. Regional cracks: Gulf Coast, Midwest, and West Coast

There is no single crack spread. The benchmark 3-2-1 is a NY Harbor futures construction, and real refineries live on regional prices that can diverge from it for months. Three regions tell the story:

• Gulf Coast (PADD 3). The center of gravity: the largest refining cluster in the world, priced off Louisiana Light Sweet crude and Gulf Coast product values, and the source of most US product exports. Gulf Coast cracks track the export market, so when foreign buyers bid aggressively for diesel, as in 2026, Gulf Coast margins capture it first. Spring-2026 industry margin surveys showed Gulf Coast 3:2:1 spreads near $50 per barrel against roughly the low $20s a year earlier.
• Midwest (PADD 2). Landlocked, supplied by pipeline and by Canadian crude, with product prices set in Chicago and Group 3 (Oklahoma/Kansas). Midwest cracks are the most volatile of the three because the region is a logistics island: a single large refinery outage or pipeline constraint with no quick resupply route can send local product prices, and local cracks, sharply higher while the coasts barely move. Spring-2026 surveys put Midwest 3:2:1 spreads in the mid-$20s, well below the Gulf Coast.
• West Coast (PADD 5). Structurally the highest-margin region in the country, and structurally the most fragile. California's CARB reformulated gasoline is a boutique specification that cannot be supplied by pipeline from other regions, so the West Coast is an island market that must balance itself. When a local refinery has problems, there is no cavalry. The standard quote is the 5:3:2 on CARBOB, and spring-2026 industry data put it around $76 to $80 per barrel, roughly triple the Midwest 3:2:1. The retail symptom of the same geography: in early September 2026, California diesel averaged $7.218 a gallon while Gulf Coast diesel averaged $5.360, a regional gap of almost $1.90 that is highly unusual.

The same logic applies internationally. Europe's diesel premium over Brent widened from about $21 to $100 per barrel during the 2026 squeeze, lifting wholesale diesel toward $194 per barrel, which shows the distillate shortage was an Atlantic Basin phenomenon, not just an American one. When you see a crack spread quoted without a region attached, assume it is the NY Harbor futures benchmark and remember that your local refiner may be living in a different market.

8. What a negative crack means

A crack spread can go below zero, and it means exactly what the arithmetic says: the products are worth less than the crude it took to make them. The refiner loses money on every incremental barrel processed, before operating costs are even counted. It is the market's way of shouting that there is too much refining capacity running for the available demand.

The textbook case is spring 2020. As COVID-19 lockdowns crushed driving, US gasoline demand collapsed from 9.7 million barrels per day in the week ending March 20 to a record low of 5.1 million in the week ending April 3. Gasoline prices fell faster than crude prices, and the gasoline crack spread fell to negative 8 cents per gallon, its lowest since January 2019. RBOB cracks on NYMEX settled below $1 per barrel on March 16, and regional gasoline cracks around the country moved into negative territory. Diesel cracks, notably, stayed strong near $16 per barrel, because freight and farming kept burning diesel while passenger cars sat parked.

Negative cracks are self-correcting, but the correction is painful. Refiners cut runs, idle units, and in 2020 permanently closed several plants. Each closure removes supply until the margin recovers enough for the survivors. That is the economic restoring force behind the spread's tendency to mean-revert: sustained wide margins pull runs up and add product supply, sustained narrow margins shut capacity in. But "mean reversion" here is not a gentle statistical drift. It works through bankruptcies, layoffs, and permanent closures on the way down, and through record pump prices on the way up. The mechanism is real; the human cost of each swing is also real.

9. Honest limits: what the spread cannot tell you

The crack spread is one of the best diagnostic numbers in energy markets, and it is routinely overinterpreted. Six limits, stated plainly:

• It is a margin signal, not a price forecast. The spread can widen while crude prices fall, which means cheaper oil and record refiner profits at the same time. Knowing the crack tells you the direction of the margin. The direction of the flat price needs the crude leg too, and crude has its own drivers. Our explainer on WTI vs Brent spreads covers the crude side of that equation.
• It is gross, not net. The spread subtracts feedstock and nothing else. Two refineries facing the same crack can have very different profits depending on energy costs, maintenance backlogs, debt loads, and regulatory compliance costs such as renewable fuel obligations. Company margins and market cracks rhyme; they are not the same number.
• It is a benchmark, not your refinery. The quoted 3-2-1 uses WTI and NY Harbor futures. A refinery running discounted heavy Canadian crude in the Midwest, or one configured to maximize diesel in Europe, lives on a different spread. Basis differentials between the benchmark and the actual barrels can be tens of dollars.
• Blended spreads hide single-product blowouts. September 2026 is the exhibit: the 3-2-1 sat 15 percent below its record while the diesel crack set records. Anyone watching only the headline number missed the actual shortage. When one product is dislocated, read the single-product cracks, not the blend.
• Mean reversion is not a schedule. The restoring force is real, but it operates through physical events: turnarounds ending, new capacity starting, demand destruction, recessions. "Wide spreads attract supply" can take quarters to work, and policy interventions, export bans, and wars can hold spreads wide far longer than the historical average suggests.
• The curve matters, not just the prompt quote. Headlines quote the front-month spread. Refiners hedge and plan on calendar strips months out. A wide prompt spread with a weak forward curve says the market expects the dislocation to be temporary; a wide curve says it expects the tightness to persist. Always check which one you are looking at.

10. A reader's checklist

When a crack spread headline crosses your screen, run these six checks before drawing conclusions:

1. Which spread, which region? 3-2-1, 2-1-1, or 5-3-2; NY Harbor futures, Gulf Coast, Midwest, or West Coast. An unattributed number is usually the NYMEX 3-2-1.
2. Where is it against the bands? $10 to $20 is healthy, $20 to $30 elevated, $30 to $40 stress, above $40 acute stress. In 2026, "above $40" has been the neighborhood, not the exception.
3. Which leg is moving? Check the gasoline and distillate cracks separately. If they are diverging, the story is in the split, not the average.
4. Crude shock or product shock? Crude up and cracks compressing means a crude problem. Cracks widening means a refining or product problem. The investment and policy implications are opposite.
5. What is the inventory backdrop? Wide cracks on top of 30-year-low distillate stocks mean something very different from wide cracks with full tanks. Stocks tell you how long the tightness can run.
6. What ends it? Turnaround season ending, a refinery restart, an export ban lifting, demand destruction. Name the specific mechanism that would narrow the spread, and watch for it. If you cannot name one, be cautious about assuming reversion.

For the consumer side of the same margin story, see our deep dive on rockets and feathers in gasoline prices, and track what drivers are actually paying on our gas prices page.

11. Frequently asked questions

Is a high crack spread good or bad?

It depends on whose wallet you mean. For refiners and their shareholders, wide spreads are excellent: margins expand and profits follow, as the second quarter of 2026 showed. For drivers, truckers, airlines, and anyone buying goods moved by diesel, wide spreads are a tax on everything, because the processing margin sits inside the pump price. For the economy, a sustained wide spread signals that the fuel-making system is running with no spare capacity, which is fragile regardless of who profits from it.

Why can gasoline be expensive when crude oil is cheap?

Because the pump price is crude plus the crack spread plus taxes and distribution, and the crack spread has a vote. If refineries are down, inventories are low, or it is driving season, products can stay expensive even as crude falls. The spread is the wedge between the wellhead and the pump, and in tight refining markets that wedge does most of the talking.

What is the difference between a crack spread and a refinery margin?

The crack spread is a market-quoted gross margin proxy: product futures minus crude futures. A refinery margin, as reported in earnings, is the company's realized number after its actual crude slate, actual product sales, operating costs, and regional basis. The two move together closely enough that executives cite market cracks on earnings calls, but the company's number is always its own.

Do crack spreads predict crude oil prices?

Not reliably. They predict refining profitability and, loosely, the direction of product prices relative to crude. A widening spread says products are likely to stay firm or crude is likely to soften relative to products; it says nothing precise about the absolute oil price next month. Treat it as a margin diagnostic, not a crystal ball.

What moves the spread the fastest?

Unplanned refinery outages and hurricanes move it in hours, because they remove physical supply with no warning. Turnaround seasons and driving season move it over weeks and are partly priced in advance. Structural forces, like permanent refinery closures or export bans, move it over months and hold it there. Speed of the move usually tells you which cause you are looking at.

Where can I follow current crack spread levels?

The EIA publishes weekly petroleum data, including regional product prices, inventories, and refinery utilization, which are the fundamentals behind the spread. Futures market data vendors quote the 3-2-1, 2-1-1, and single-product cracks daily. For context on what those moves mean for drivers, our gas prices page tracks retail prices alongside the market background.