What is the carbon footprint of steel?
About 2.2 tonnes of CO₂e per tonne of crude steel, the 2024 global average (worldsteel, scopes 1–3). The production route decides it: about 2.7 t via blast furnace, 1.7 t via gas-based direct reduction, and 0.7 t when scrap is melted in an electric arc furnace.
Steel is one material with two very different carbon stories. Made from iron ore with coal in a blast furnace, a tonne carries well over two tonnes of greenhouse gas. Melted from scrap with electricity, the same grade of steel can carry less than a third of that. Any single “carbon footprint of steel” hides that spread, so this page gives the global average first, then the figure for each route, then what the product databases say per kilogram — and why the numbers you find elsewhere rarely agree.
The footprint by production route
The World Steel Association collects emissions data from its members every year using one method across all three main routes. Its 2024 figures cover 93 companies and associations representing about half of global crude steel production.
| Route | Scrap share of input | t CO₂ per t steel | t CO₂e per t steel (GHG) |
|---|---|---|---|
| Blast furnace – basic oxygen furnace (BF-BOF) | About 10% | 2.34 | 2.67 |
| Direct reduced iron – electric arc furnace (DRI-EAF) | Under 30% | 1.47 | 1.66 |
| Scrap – electric arc furnace (scrap-EAF) | Over 70% | 0.69 | 0.71 |
| Global average, weighted by route | — | 1.92 | 2.18 |
These are scope 1, 2 and upstream scope 3 figures per tonne of crude steel — the liquid steel cast at the plant. Rolling, coating and fabricating it into a finished product adds more. worldsteel notes that the GHG indicator, first collected for 2024, reads higher than the CO₂-only one because of the wider boundary, not because emissions rose.
Why the route matters so much
Most of steel’s footprint is chemistry, not fuel choice. Iron ore is iron oxide. To make iron, the oxygen has to be stripped off, and in a blast furnace that is done with carbon from coke: the oxygen leaves as CO₂. Those are process emissions, and they cannot be removed by switching to renewable electricity.
- BF-BOF reduces ore with coke in a blast furnace, then refines the iron to steel in a basic oxygen furnace. It is the dominant route worldwide and the most carbon-intensive.
- DRI-EAF reduces ore with natural gas (or, in new plants, hydrogen) without melting it, then melts the iron in an electric furnace. Gas-based reduction releases less CO₂ than coke; hydrogen-based reduction releases water instead.
- Scrap-EAF skips ore reduction altogether. It melts existing steel with electricity, so its footprint is mostly the electricity and depends on the grid supplying it.
The IPCC’s default Tier 1 factors for national inventories show the same split for the steelmaking step alone. In the IPCC 2006 Guidelines, basic oxygen furnace steel is assigned 1.46 t CO₂ per tonne [GreenCalculus metals.iron_steel.bof · IPCC VOL3 2006 Vol 3 Ch 4 Table 4.1 · v2026.237], blast-furnace pig iron 1.35 t CO₂ per tonne [GreenCalculus metals.iron_steel.blast_furnace_pig_iron · IPCC VOL3 2006 Vol 3 Ch 4 Table 4.1], and electric arc furnace steel 0.08 t CO₂ per tonne [GreenCalculus metals.iron_steel.eaf · IPCC VOL3 2006 Vol 3 Ch 4 Table 4.1]. The EAF figure is process CO₂ only; the electricity is counted separately. These are the factors behind the iron and steel process calculator and the process methodology.
Per kilogram of product: what the EPD data says
Buyers rarely buy crude steel. They buy rebar, beams, sheet and tube, and the figure that matters for a building or a product is cradle-to-gate per kilogram of that product — modules A1–A3 in an environmental product declaration. The German Ökobaudat database publishes generic values for common steel products, and the spread is wide:
| Product | kg CO₂e per kg | Typical route for this product type |
|---|---|---|
| Reinforcing bar (rebar) | 0.47 | Mostly scrap-EAF |
| Hot-rolled sections | 0.56 | Mostly scrap-EAF |
| Galvanised steel | 2.48 | Mostly BF-BOF, plus zinc coating |
| Hot-rolled plate and sheet, 2–20 mm | 2.74 | Mostly BF-BOF |
| Stainless steel fasteners (screws) | 6.88 | EAF, plus nickel and chromium alloying |
A kilogram of CO₂e per kilogram of steel is the same ratio as a tonne per tonne, so these sit directly beside the worldsteel figures. Long products such as rebar and sections are usually made from scrap in electric furnaces, which is why they come in far below the global average. Flat products such as hot-rolled plate are mostly made from ore, and stainless steel carries the heavy footprint of its alloying metals — the one stainless dataset here is for screws, so treat it as indicative for stainless products generally. The steel and aluminium embodied-carbon calculator applies these factors to a bill of quantities.
“Steel” is not one number. Rebar from a European scrap mill and hot-rolled plate from a blast furnace differ by a factor of about six per kilogram.
Why published figures disagree
Search for the carbon footprint of steel and you will find figures from under 0.5 to over 3 tonnes per tonne. Most of the difference is not error; it is that each source answers a different question. Five choices explain nearly all of it:
| Choice | Lower figure | Higher figure |
|---|---|---|
| Route and scrap share | Scrap-EAF, high recycled content | BF-BOF, ore-based |
| Boundary | Direct (scope 1) emissions at the plant only | Scopes 1, 2 and upstream 3 (cradle to gate) |
| Gases | CO₂ only | CO₂e including methane and nitrous oxide |
| Product stage | Crude steel at the caster | Finished, coated or fabricated product |
| Electricity | Low-carbon grid or renewable contract | Coal-heavy grid |
Recycling credits add a sixth source of confusion. Some EPDs report a “module D” benefit for steel that will be recycled at end of life; that credit sits outside the A1–A3 figure and should never be subtracted from it. The EU’s carbon border adjustment mechanism uses yet another boundary — its embedded emissions rules count direct emissions and, for some goods, electricity, under defined production-process boundaries — so a CBAM figure and an EPD figure for the same coil are not comparable without adjustment. Why emission factors disagree works through the same axes for other materials.
Steel in the global total
The world produced 1,886 million tonnes of steel in 2024. worldsteel puts the sector’s total emissions at around 4.1 billion tonnes of CO₂e — about three-quarters of it direct emissions — or 7–8% of global anthropogenic greenhouse-gas emissions. That puts steel among the largest industrial sources of emissions in the world.
For most companies, steel shows up not in their own operations but in their supply chain, as scope 3 category 1, purchased goods. For construction, automotive and machinery companies it is often one of the largest items in the inventory, and one where the choice of supplier and product changes the figure more than any other decision.
Worked example: the steel in a building frame
A mid-rise building uses 200 tonnes of reinforcing bar and 150 tonnes of hot-rolled sections. The first calculation uses the Ökobaudat A1–A3 factors above as they read at the time of writing; the second applies the worldsteel 2024 global average to the same tonnage. Illustrative.
| Item | Product-specific (Ökobaudat) | Global average (2.18 t/t) |
|---|---|---|
| Rebar, 200 t | 200,000 kg × 0.47398 = 94.8 tCO₂e | 200 t × 2.18 = 436.0 tCO₂e |
| Hot-rolled sections, 150 t | 150,000 kg × 0.56029 = 84.0 tCO₂e | 150 t × 2.18 = 327.0 tCO₂e |
| Total | 178.8 tCO₂e | 763.0 tCO₂e |
The same 350 tonnes of steel comes to 178.8 or 763.0 tCO₂e — a fourfold difference — depending on whether the factor reflects the actual product and route or a global crude-steel average. Neither is wrong; they answer different questions. For a building’s embodied carbon, use the product’s own EPD where you have one, the product-type generic value where you do not, and the global average only as a last resort.
For the concrete around the reinforcing bar, see the carbon footprint of concrete.
How the footprint comes down
Four levers cut steel’s footprint, and they are not interchangeable:
- More scrap in electric furnaces. The cheapest large cut today, limited by how much scrap exists. Scrap supply is finite, so specifying high recycled content on one project partly moves low-carbon scrap away from another rather than creating more of it.
- Cleaner electricity for EAFs. A scrap-EAF on a low-carbon grid can fall well below the 0.71 t average; on a coal-heavy grid it rises.
- New ore-based routes. Hydrogen-based direct reduction, and carbon capture on blast furnaces, address the process emissions that scrap cannot reach. Ore-based steel will still be needed because there is not enough scrap to meet demand.
- Using less steel. Lighter designs, higher-strength grades and longer service lives reduce tonnage, which is the one lever that works on every route.
For a buyer, the practical step is to ask suppliers for a product-specific EPD under EN 15804 or ISO 14067, stating the route, the scrap share and the electricity source. A figure without those three facts cannot be compared with any other.
Frequently asked questions
On worldsteel’s 2024 global average, 1.92 tonnes of CO₂, or 2.18 tonnes of CO₂e when methane, nitrous oxide and upstream mining are included, per tonne of crude steel. Blast-furnace steel is about 2.3–2.7 tonnes; steel melted from scrap in an electric arc furnace is about 0.7 tonnes.
Yes, per tonne. Melting scrap in an electric arc furnace avoids reducing iron ore, the step that causes most of steel’s emissions, so scrap-based steel averages about 0.7 tonnes of CO₂e per tonne against about 2.7 for blast-furnace steel. Because scrap supply is limited, though, global emissions fall only as fast as scrap availability and cleaner ore-based routes allow.
About 7–8% of global anthropogenic greenhouse-gas emissions, according to worldsteel. The sector emitted around 4.1 billion tonnes of CO₂e in 2024 while producing 1,886 million tonnes of steel, with about three-quarters of those emissions released directly at steel plants.
Use the supplier’s product-specific EPD if you have one. Otherwise use a generic cradle-to-gate value for the product type, such as the Ökobaudat or ICE values for rebar, sections or sheet. Use a global crude-steel average only when you know nothing about the product, and state the boundary, gases and source you used.
About 2.2 tonnes of CO₂e per tonne of crude steel on the 2024 global average, covering scopes 1, 2 and upstream scope 3. It ranges from about 0.7 tonnes for scrap-based electric arc furnace steel to about 2.7 tonnes for blast-furnace steel, and finished products such as stainless steel or coated sheet can be higher still.