Are Seed Oils Bad for You? Ask How They're Made, Not What One Blood Marker Did

Start with a question no study in this debate has ever asked: what does it take to get the oil out?

Olive oil is crushed fruit. You press it, and what runs out is extra-virgin olive oil — mechanically separated juice, nothing added, nothing stripped. Coconut oil is pressed meat. Butter is cream agitated until the fat globules cluster; you can make it in a jar by shaking it, and people did exactly that in wooden churns for thousands of years.

Canola oil cannot be made that way, because a seed will not give up its oil to a press alone. The commodity process runs: clean the seed, crack it, flake it, heat-condition it, press what will come, then wash the rest out of the remaining meal with hexane — a petroleum solvent — evaporate and recover the solvent, degum, neutralize with alkali, bleach with clay, and finally deodorize by steam-stripping at roughly 200°C, because what comes out before that step does not taste or smell like food.

Both end in a bottle of fat. They are not the same kind of object, and nothing about that difference shows up on a cholesterol panel.

We don't cook with seed oils. That is our position, and the rest of this article is about why it does not rest on the claim you have probably heard.

This article is educational, not medical or nutrition advice. Every figure comes from a named peer-reviewed source.

Key statistics at a glance

• Rapeseed was an industrial lubricant before it was a food. It clung to metal under the heat and steam of marine engines, which is why it mattered in the Second World War — documented agricultural history
• Traditional rapeseed ran 40% or more erucic acid, linked to heart damage in animal studies; canola is the cultivar bred to remove it, released in 1974. The name is an abbreviation: Canada, Oil, Low Acid — Stefansson & Downey, University of Manitoba
• Polyunsaturated fats are the most oxidation-prone fats there are — a direct consequence of having more double bonds — and a bottle's oxidation state appears nowhere on its label
• A systematic review of randomized trials found no causal relationship between linoleic acid intake and inflammatory markers — Johnson & Fritsche, Journal of the Academy of Nutrition and Dietetics, 2012
• Ultra-processed foods supply 53.0% of US adult calories, and fried and packaged foods are the dominant route by which seed oils are eaten — CDC/NCHS Data Brief 536, 2025

The processing gap, laid out

This is the comparison that decides the question for us, and it is the one no trial has ever run.

Read what each of those nine stages is doing. Pressing gets some of the oil. Hexane gets the rest — and hexane is there because mechanical pressing alone leaves too much oil in the meal to be commercially worth it. Degumming, neutralizing, bleaching and deodorizing all exist to remove things the earlier steps produced or exposed: free fatty acids, phospholipids, pigments, and the smell and taste of crude seed oil.

An extra-virgin olive oil that went through that chain would no longer be extra-virgin. It would be a different product, and the label would have to say so.

By the NOVA classification the rest of our reporting relies on, these are not the same category of food. It would be incoherent to argue that industrial processing is the central problem in the food supply — which the controlled feeding evidence supports — and then set that aside the moment the processed item scores well on a blood panel.

The claim we're not going to make

Here is where we part company with most people writing on our side of this.

The specific claim that omega-6 linoleic acid drives inflammation is not supported by the controlled evidence, and we are not going to repeat it. A systematic review of randomized trials in healthy people found no causal relationship between linoleic acid intake and inflammatory markers (Johnson & Fritsche, 2012). A 2017 meta-analysis pooling 30 randomized studies found no significant effect on IL-6 or CRP. Researchers went looking for the mechanism and it did not appear in people.

The biochemical story is genuinely appealing — linoleic acid converts to arachidonic acid, arachidonic acid is a precursor to inflammatory eicosanoids — and every step of it is real. What it leaves out is that the conversion is tightly regulated and inefficient, so raising intake does not reliably raise tissue arachidonic acid, and that the eicosanoids downstream include anti-inflammatory and resolving mediators alongside the inflammatory ones.

We are saying this plainly for a practical reason. If our argument were "seed oils cause inflammation," the first reader who checks finds trials saying otherwise and discards everything else on this site. The case against industrially refined oil does not need a mechanism that isn't there — and a position that survives being checked is worth more than one that sounds better.

What the studies actually measured

Now the other half, which is the half that usually goes missing.

A study is a question with a method attached. Translate the seed oil trials out of journal language and the question is:

Does raising one fatty acid, over a few weeks, in participants eating an otherwise controlled diet, move one or two proteins in the blood?

The answer is no. That is a real answer to a small question. Here is what it is not:

• It is not a statement that the oil is good food. A blood marker says nothing about processing, about what the oil displaces, or about what else it does or doesn't carry.
• It is not an outcome. LDL and CRP are surrogate markers — stand-ins used because waiting for heart attacks takes decades. Surrogates have misled badly before. The last time nutrition science decided a fat was healthy on the strength of one panel, it gave us margarine and partially hydrogenated oil, and the FDA spent 2015–2018 pulling it back out of the food supply.
• It is not a test of how much. Substitution trials hold calories constant. Someone eating an enormous amount of butter and someone eating a little are not the same case, and "use less butter" is a change these trials never tested.
• It is not a test of the product you can buy. Which is the next section.

There is a particular version of this worth naming, because it gets cited as though it settles things. A 2019 pooled analysis of 30 cohorts found higher blood levels of linoleic acid associated with lower cardiovascular mortality. Measuring blood rather than relying on food questionnaires is a real methodological strength. But circulating linoleic acid is not a record of how much refined oil someone poured into a pan — it tracks whole dietary patterns, including nuts, seeds and whole foods that carry linoleic acid inside an intact food matrix. Reading a cohort of blood levels as a verdict on commodity oil is a substitution the study never made.

"Seed oil" is a category, not a substance

A trial specifies its oil. It names the type, sources it, stores it to protocol, and uses it fresh. What it produces is a finding about that product under those conditions.

The category on a supermarket shelf has none of that:

• Extraction. Cold-pressed and expeller-pressed oils never meet a solvent. Most commodity oil is hexane-extracted, then bleached and deodorized. Same botanical seed, materially different product.
• Oxidation state. How oxidized a given bottle is depends on refining temperature, light exposure, oxygen, time on the shelf, and how long it sat in a warm truck. None of that appears on the label, and two bottles of the same brand can differ.
• Reuse. Oil held hot in a commercial fryer for days is chemically not the oil that was poured in — and that is how most people actually consume it.
• Blends. "Vegetable oil" is frequently a blend whose composition shifts with commodity prices between production runs.

So a trial finding no effect from a defined, fresh, well-handled oil has told you something true about that oil. It has told you nothing about a bottle that spent eight months under fluorescent light, or a fryer running since Tuesday. Those questions were not asked, so they were not answered.

Two consequences, and we are stating both. The confident dismissal — "trials show seed oils are fine" — claims more than the trials support; they cannot speak to a category this heterogeneous. And cutting the other way: we cannot assert that supermarket oil is worse than trial oil, because nobody has properly checked. We are not going to invent a finding to fill that gap. What we will say is that when a category varies this much and the label discloses almost none of it, a shopper has no way to know what they are buying — the same disclosure problem that runs through the added-sugar gap.

There is also a trend worth naming without overstating: refining has become more industrialized as volume has grown. We have not found data quantifying how the average retail bottle has changed over time, so we flag it as an open question rather than a claim.

The two concerns that do survive

Neither is about omega-6 as such.

Repeatedly heated oil is a different substance. Frying oil held at high temperature and reused undergoes oxidation and polymerization, generating aldehydes and other degradation products not present in fresh oil. This is straightforward chemistry, and it is a real reason to distinguish the bottle in your kitchen from the oil in a fryer that has run all week. Note this is an argument about thermal degradation, which applies to fats generally.

The company they keep. Seed oils are the fat in most fried fast food, packaged snacks and commercial bakery products. Someone who cuts seed oils usually cuts those foods, feels better, and credits the oil. The confound is enormous — which is exactly why the ultra-processed food evidence sits on much firmer ground than the seed oil evidence, and why we lead with it.

What we use instead, and why

Extra-virgin olive oil and unrefined coconut oil. Organic where possible.

Not because a trial ranked them first on a biomarker. Because they are foods you could make in a kitchen, from a plant you could hold, with nothing added and nothing stripped out. That standard is older than nutritional epidemiology and it does not need a blood panel to justify it.

Butter is churned cream, and it is fine. We are not going to rank butter against canola on a single number, because compressing a multi-variable question into whichever variable supports your conclusion is the error this whole article is about. Butter carries fat-soluble vitamins A, D and K2 in a dairy matrix whose effects are not reducible to its fatty acid profile; refined oil is close to pure fat. Anyone declaring a clean winner in either direction is doing the thing we just spent 1,500 words describing.

If you want the practical version: use real fat, use less of it, and cook more of your own food. None of that depends on who wins the omega-6 argument.

Frequently asked questions

Do seed oils cause inflammation?

Not by the mechanism usually claimed. A systematic review of randomized trials found no causal relationship between linoleic acid intake and inflammatory markers, and a 2017 meta-analysis of 30 studies found no significant effect on IL-6 or CRP. We think there are better reasons to avoid them, and repeating a claim the evidence does not carry costs you credibility on the claims that hold.

So why avoid them?

Because of what they are and how they are made. Commodity seed oil requires solvent extraction with hexane, bleaching, and deodorizing at around 200°C. Olive oil requires a press. No trial in this literature was designed to isolate the effect of that refining chain, so no trial in this literature has anything to say about it — and "nobody tested it" is not the same sentence as "it's fine."

Is canola oil really made from an industrial lubricant?

The history is real and worth knowing, though it is often told sloppily. Rapeseed oil was genuinely used as a machine lubricant, valued because it clung to metal under heat and steam. Traditional rapeseed ran 40% or more erucic acid, linked to heart damage in animal studies. Canola is the low-erucic cultivar bred at the University of Manitoba to remove it, released in 1974 — the name stands for Canada Oil Low Acid. That is context, not an argument that canola is poison.

Are all seed oils the same?

No, and that undercuts every confident claim in this debate. Cold-pressed oil never meets a solvent; most commodity oil is hexane-extracted, bleached and deodorized. Oxidation varies with refining temperature, light, oxygen and shelf time, none of it disclosed on the label. Trials test one defined product handled to protocol — not the bottle that has been under supermarket lighting for eight months.

Is reused frying oil different from fresh oil?

Yes. Oil held at high heat and reused undergoes oxidation and polymerization, producing degradation compounds absent from fresh oil. That is a concern about thermal degradation, which applies to fats in general rather than to omega-6 specifically.

Is butter better than seed oil?

We are not going to rank them on a single number. The trials substituting polyunsaturated for saturated fat show improved blood lipid markers — one measurement, and a surrogate rather than an outcome. Deciding which is better food on that basis ignores processing, what else is in there, how much anyone eats, and what the rest of the meal looks like. Butter is churned cream; refined seed oil is an industrial extract. Weigh that how you like, but weigh more than one variable.

Sources

• PubMed — Johnson & Fritsche, Effect of Dietary Linoleic Acid on Markers of Inflammation in Healthy Persons: A Systematic Review of Randomized Controlled Trials (2012)
• Nutrients — Lai, Ryder, Tintle, Jackson, Kris-Etherton & Harris, Red Blood Cell Omega-6 Fatty Acids and Biomarkers of Inflammation in the Framingham Offspring Study (2025)
• Cell Metabolism — Hall et al., Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain (2019)
• CDC/NCHS — Data Brief 536, Ultra-processed Food Consumption in Youth and Adults (2025)
• FDA — Final Determination Regarding Partially Hydrogenated Oils (2015)

Related: Do Ultra-Processed Foods Make You Eat More? — the evidence that actually supports eating less packaged food.