Few nutrition topics have become as polarized as seed oils.
So are they harmful? Are they safe?
Or is the answer more nuanced than either side wants to admit?
Critics describe soybean, corn, sunflower, safflower, cottonseed, grapeseed, and canola oils as major drivers of inflammation, obesity, mitochondrial dysfunction, and chronic disease.
Defenders of seed oils point to their ability to lower LDL cholesterol and to observational studies associating higher linoleic-acid levels with favorable cardiovascular and metabolic outcomes.
So, what’s the truth?
In this video, we’re going to do a deep dive into the literature.
You’ll learn what’s real and what’s not.
You’ll leave with either a deep appreciation of the complexity, or perhaps just more confused than ever. Either way, I promise you’ll learn something you didn’t already know.
So, let’s jump right in.
The term seed oil identifies where an oil came from.
It does not tell us its exact fatty-acid composition, how easily it oxidizes, how it was processed, or how it was used.
Traditional soybean, corn, sunflower, safflower, cottonseed, and grapeseed oils contain substantial amounts of linoleic acid, an omega-6 polyunsaturated fatty acid.
Canola oil contains less linoleic acid, more monounsaturated oleic acid, and some alpha-linolenic acid, an omega-3 fatty acid.
High-oleic sunflower and safflower oils contain far less linoleic acid than their conventional counterparts.
So, these products should not be treated as interchangeable.
The first thing we need to know is that the number of double bonds in a fatty acid influences its susceptibility to oxidation; the more double bonds present, the more susceptible.
Linoleic acid contains two double bonds. Oleic acid, the predominant fat in olive oil, contains one. Saturated fatty acids contain none.
Linoleic acid is therefore more susceptible to lipid peroxidation than oleic acid or saturated fat.
Lipid peroxidation is the oxidative degradation of polyunsaturated fats, producing reactive compounds that can spread through membranes and damage membranes, proteins, DNA, lipoproteins, and mitochondria.
That chemical difference is established. Up to this point, nothing I’ve said is controversial.
The controversy concerns how much oxidation occurs under real-world conditions and whether the resulting exposure causes meaningful long-term damage.
One thing worth mentioning is that linoleic acid is an essential fatty acid.
Humans cannot synthesize it, and severe deficiency impairs skin-barrier function and other physiological processes.
The relevant question is not whether the diet should contain any linoleic acid (it needs to). It is whether refined oils should supply large amounts of it every day when essential requirements can be met through a varied diet.
Seed-oil debates often become confused because different forms of evidence answer different questions.
Chemical and mechanistic evidence shows what reactions can occur and how they could affect cells.
Biomarker studies show whether a pathway changes in humans.
Short-term feeding trials measure near-term physiological effects under controlled conditions.
Observational studies identify long-term associations but cannot fully establish cause and effect.
Randomized clinical outcome trials provide the strongest evidence of disease benefit or harm, but high-quality decades-long trials of seed-oil exposure are largely unavailable.
No single level settles the entire question.
A mechanism cannot prove that a food causes chronic disease. But a short-term reduction in LDL or an unchanged CRP level cannot establish that decades of greater tissue linoleic-acid exposure are harmless to mitochondria.
The quality of the conclusion depends on matching the evidence to the question being asked.
The seed-oil concern becomes easier to understand when three separate stages are distinguished.
Oil can oxidize during processing, storage, transportation, restaurant use, or repeated high-temperature cooking. This creates compounds that are already present when the food is eaten.
Unoxidized dietary linoleic acid can enter circulating lipids, adipose tissue, lipoproteins, and cellular membranes.
Once incorporated, tissue linoleic acid remains available for enzymatic oxidation and free-radical lipid peroxidation inside the body.
The evidence is strongest that repeated heating creates oxidation products. It is also established that dietary linoleic acid changes human tissue composition.
The unresolved question is whether endogenous oxidation of a larger tissue linoleic-acid pool produces meaningful cumulative damage over decades.
When linoleic-acid intake rises, the proportion of linoleic acid in adipose tissue and several circulating lipid pools also rises.
An analysis of published US adipose-tissue studies estimated that linoleic acid increased from approximately 9.1% to 21.5% of the fatty acids measured in subcutaneous body fat between 1959 and 2008.
This does not mean linoleic acid accumulates indefinitely. Fatty acids are continuously released, oxidized, and replaced, and tissue composition eventually reaches a new equilibrium.
It does establish that changes in dietary intake can substantially alter the fatty-acid composition of the human body.
That matters because adipose tissue is not an inert storage container. Stored fatty acids are continually released into circulation and can be incorporated into lipoproteins, cellular membranes, and signaling molecules.
A larger tissue pool of linoleic acid therefore creates a larger supply of oxidizable substrate.
The historical rise in tissue linoleic acid does not prove that it caused the parallel rise in obesity, diabetes, or other chronic diseases. Many dietary, environmental, and behavioral variables changed over the same period.
But the first step in the proposed long-term pathway is established:
Higher dietary linoleic acid changes human tissue composition.
The central mechanistic concern can be summarized as follows:
Higher long-term LA intake
→ greater tissue and membrane LA
→ a larger oxidizable substrate pool
→ enzymatic and free-radical lipid peroxidation
→ OXLAMs, lipid hydroperoxides, reactive aldehydes, and oxidized phospholipids
→ cumulative mitochondrial, vascular, or cellular stress
Linoleic acid can be oxidized enzymatically through pathways involving lipoxygenases, cyclooxygenases, and cytochrome P450 enzymes. It can also undergo nonenzymatic free-radical peroxidation.
These reactions generate a broad family of oxidized linoleic-acid metabolites, commonly called OXLAMs. They include:
9- and 13-HODE
9- and 13-oxoODE
EpOMEs
DiHOMEs
Linoleic-acid peroxidation can also produce lipid hydroperoxides and downstream reactive aldehydes.
Some OXLAMs can alter mitochondrial and cellular signaling, while downstream lipid-peroxidation products such as reactive aldehydes can damage proteins, lipoproteins, DNA, and cellular membranes.
More tissue linoleic acid provides more substrate from which these products can be formed.
But OXLAMs are not one uniform toxin. Their effects depend on the specific compound, concentration, stereochemistry, tissue, receptor, and physiological context. Some participate in normal adaptation and signaling.
Circulating OXLAM concentrations also do not necessarily reflect exposure within a specific tissue. No single plasma OXLAM has been validated as a reliable measure of seed-oil-induced mitochondrial or tissue damage.
The concern is therefore not simply that OXLAMs exist. It is whether chronic tissue enrichment changes their quantity, balance, location, or persistence in a harmful direction.
A 12-week dietary intervention in people with chronic headaches examined what happened when linoleic acid intake was reduced.
The intervention lowered linoleic acid in multiple circulating lipid fractions and reduced four measured metabolites: 9-HODE, 13-HODE, 9-oxoODE, and 13-oxoODE.
A separate controlled feeding study compared diets rich in linoleic acid or oleic acid.
During the high-LA intervention, urinary 8-iso-PGF2α, a marker of lipid peroxidation, increased, while nitric oxide metabolites declined. Some direct comparisons between the high-LA and high-oleic groups were not statistically significant, so the result was suggestive rather than definitive.
Together, these studies establish that dietary linoleic acid can alter circulating OXLAM concentrations and at least one marker of lipid peroxidation in humans.
They do not establish that these biochemical changes cause chronic disease.
The first study could not determine whether OXLAM concentrations fell because of reduced production, lower ingestion of preformed oxidation products, altered transport, or faster clearance. Neither study measured long-term mitochondrial function or hard clinical outcomes.
The TL;DR is that we know the pathway responds to diet.
But its long-term clinical importance remains unresolved.
The consequences of a larger oxidizable substrate pool may not be uniform across the population.
Lipid peroxidation is influenced by more than the quantity of linoleic acid present. Relevant factors include:
Iron and other catalytic metals
Mitochondrial reactive-oxygen species production
Antioxidant capacity
Aldehyde-detoxification systems
Chronic inflammation
Insulin resistance
Liver function
Alcohol exposure
Tissue oxygenation
The rate at which oxidation products are cleared
This creates a plausible susceptible-subgroup hypothesis.
People with chronic inflammation, metabolic syndrome, iron overload, liver disease, impaired antioxidant defenses, or preexisting mitochondrial stress could theoretically produce more oxidized lipids or clear their reactive products less effectively.
To be clear, this is a mechanistically coherent hypothesis, but direct human evidence is limited. Current studies have not established which groups, if any, experience greater harm from a given linoleic-acid intake.
This is an important nuance because population averages can obscure meaningful effects in vulnerable subgroups. In other words, if a susceptible subgroup is small and is not analyzed separately, a null average effect across the full population can obscure a meaningful effect within that subgroup.
Cardiolipin is a specialized phospholipid concentrated in the inner mitochondrial membrane. It helps organize respiratory-chain complexes, stabilize membrane structure, and support oxidative phosphorylation.
Oxidation of cardiolipin can impair electron transport and participate in apoptosis and inflammatory signaling. Cardiolipin oxidation is therefore one plausible route through which lipid peroxidation could damage mitochondria.
However, linoleic acid itself is not an abnormal component of cardiolipin.
In healthy human heart and skeletal muscle, the predominant species is tetralinoleoyl cardiolipin, which contains four linoleic-acid chains. Loss or abnormal remodeling of this species occurs in heart failure and inherited mitochondrial disorders.
A two-week randomized trial also found that LA-rich oil increased tetralinoleoyl cardiolipin in peripheral blood mononuclear cells. The study did not determine whether mitochondrial function changed or whether the same effect occurred in heart, liver, or skeletal muscle.
Still, it reinforces an important distinction: incorporating LA into cardiolipin is not inherently harmful. The proposed concern is excessive oxidation or abnormal remodeling under vulnerable conditions.
One common claim is that linoleic acid is converted into arachidonic acid, which is then used to produce inflammatory eicosanoids.
The pathway exists, but increasing dietary linoleic acid does not produce a proportional increase in tissue arachidonic acid. Conversion is regulated and limited in humans.
Controlled trials have not shown that increasing LA within commonly studied intake ranges reliably raises CRP, interleukin-6, tumor-necrosis factor, fibrinogen, or other conventional markers of systemic inflammation.
However, the duration and sampling schedules are a limiting factor. The trials in an earlier systematic review lasted approximately two to nine weeks, while those included in a later meta-analysis generally lasted four to 24 weeks.
In most cases, inflammatory markers were assessed at baseline and at the end of the dietary period rather than continuously or repeatedly after meals.
These studies are long enough to argue against a large, sustained systemic inflammatory response emerging within weeks or months. CRP and circulating cytokines can respond relatively quickly to a persistent inflammatory stimulus.
But they were not designed to exclude a slower process involving years of tissue linoleic-acid enrichment, repeated low-level lipid peroxidation, organ-specific oxylipin signaling, cardiolipin oxidation, or cumulative mitochondrial stress.
Endpoint measurements could also miss transient post-meal changes or localized tissue effects that do not substantially alter circulating CRP.
The evidence therefore argues against describing LA as predictably and broadly inflammatory. It does not resolve the separate hypothesis of cumulative tissue oxidation.
Rodent studies suggest that high-LA diets can alter endocannabinoid signaling and promote adiposity under some experimental conditions.
Translation to humans is uncertain. Rodents convert linoleic acid to arachidonic acid more readily, and experimental diets may differ in omega-3 content, total fat, and other important variables.
Human research does not show that polyunsaturated fat is uniquely obesogenic. In some controlled overfeeding trials, excess calories from saturated fat produced more liver and visceral-fat accumulation than excess calories from polyunsaturated fat.
Seed oils may still contribute indirectly to weight gain because they are common in foods engineered for easy overconsumption: fries, chips, pastries, sauces, and packaged snacks.
Reducing these foods can improve appetite control and calorie intake. That does not establish that linoleic acid was the sole cause of the benefit.
The strongest case for avoidance applies to oils that have already undergone substantial degradation.
Repeated high-temperature heating produces lipid hydroperoxides, aldehydes, epoxides, cyclic compounds, and polymers. Some of these products enter the food being cooked.
All fats degrade under harsh conditions, but conventional high-linoleic oils are generally more susceptible to peroxidation than high-oleic or predominantly saturated fats.
Actual stability also depends on temperature, heating duration, oxygen exposure, antioxidant content, refining, metal contamination, and the condition of the oil.
The evidence can be divided into three levels:
Repeated heating produces oxidation products: established
Many of those products can damage cells and animals at sufficient exposure: well supported
The exact amount of chronic human disease caused by consuming repeatedly heated oil: not well quantified
The unfavorable chemistry is enough to justify minimizing the use of repeatedly used, rancid, or visibly degraded frying oils.
A fresh oil used briefly at home is not chemically equivalent to oil that has remained in a commercial fryer through repeated heating cycles.
Most conventional seed oils undergo several processing steps before reaching the consumer. These may include solvent extraction, degumming, neutralization, bleaching, dewaxing, and deodorization.
It would be inaccurate to describe every stage as harmful.
Refining can remove free fatty acids, phospholipids, metals, pesticide residues, pigments, peroxides, soaps, and volatile oxidation products that would otherwise reduce the oil’s safety, stability, taste, and shelf life.
The concern is that the same process can also remove protective compounds and create new contaminants.
Deodorization is the most consequential step.
The oil is exposed to steam under vacuum at high temperatures to remove odors, free fatty acids, and volatile compounds. Although this improves flavor and shelf stability, temperatures commonly exceeding 200°C can promote several unwanted reactions.
These include:
Formation of glycidyl fatty-acid esters
Formation of 3-MCPD and 2-MCPD esters
Cis-to-trans isomerization of unsaturated fatty acids
Degradation or removal of tocopherols, polyphenols, phytosterols, and squalene
Conjugation and polymerization of fatty acids and triglycerides
The extent of these changes depends on temperature, heating time, oil composition, precursor concentrations, equipment, and the quality of the crude oil.
Modern refining methods can substantially reduce contaminant formation, so concentrations vary among oils and manufacturers.
Research following oils through each refining stage consistently identifies deodorization as the primary point at which 3-MCPD esters and glycidyl esters rise.
In one study of camellia oil, crude and partially refined samples contained little or no measurable 3-MCPD ester, while concentrations increased substantially after deodorization. Lower deodorization temperatures produced less contamination.
Glycidyl esters form mainly from mono- and diacylglycerols under high-temperature conditions. During digestion, they can be hydrolyzed and release glycidol.
Glycidol is reactive and genotoxic and has been classified as probably carcinogenic to humans. The concern is therefore not that every exposure has been shown to cause cancer, but that glycidol can damage DNA, and no clearly harmless threshold can be assumed from its mechanism.
This has led regulators and manufacturers to establish limits and redesign refining processes to minimize glycidyl-ester formation.
3-MCPD esters form when partial glycerides and related lipid structures react with chloride-containing precursors during high-temperature processing. Temperature and duration are major determinants, with formation increasing rapidly under more severe deodorization conditions.
During digestion, 3-MCPD esters can release free 3-MCPD. Toxicological concerns come primarily from animal studies showing effects on the kidneys and male reproductive system at sufficient exposure. The human risk from the concentrations found in individual oils is less certain and depends on dose, frequency of consumption, body size, and the specific product.
Evidence surrounding 2-MCPD is less developed, but it is monitored alongside 3-MCPD because it forms through related processes.
These contaminants are not unique to high-linoleic seed oils. They occur across refined vegetable fats, and some of the highest historical concentrations have been found in palm oil. Their presence is therefore evidence of a high-temperature refining problem, not proof that linoleic acid itself is toxic.
Refining can also reduce compounds that normally protect an oil from oxidation.
Tocopherols, polyphenols, phytosterols, squalene, and other minor constituents can be partially removed or degraded during neutralization, bleaching, and especially deodorization. Their loss does not automatically make the finished oil dangerous, but it can leave a highly unsaturated oil with fewer of its original antioxidant defenses.
High-temperature deodorization can also cause small amounts of cis unsaturated fatty acids to isomerize into trans configurations and can promote triglyceride polymerization. The magnitude depends heavily on how aggressively the oil is processed.
Bleaching and neutralization are often described online as inherently toxic because they use caustic soda, phosphoric or citric acid, and activated clays.
These substances are processing aids rather than intended ingredients in the finished oil. Neutralization removes free fatty acids, soaps, metals, and other impurities. Bleaching earth adsorbs pigments, peroxides, trace metals, and some contaminant precursors.
These stages may reduce beneficial micronutrients, but they can also make crude oil safer and more stable. Their mere use does not demonstrate that harmful quantities of processing chemicals remain in the finished product.
Hexane is used to extract oil efficiently from some seeds. High occupational exposure is neurotoxic, but that does not establish comparable risk from food.
Hexane is volatile and is largely removed through desolventizing, distillation, and deodorization. Properly manufactured oils should contain only trace residues. Compared with glycidyl esters, 3-MCPD esters, and heat-generated oxidation products, residual hexane is a less substantiated concern.
Industrial refining is a tradeoff.
It removes substances that can make crude oil unstable or unsuitable for consumption, while high-temperature processing can create contaminants and reduce naturally protective compounds. The main concern is not that an oil has touched a chemical during processing. It is the combination of:
High deodorization temperature
Long processing time
Contaminant precursors in the crude oil
Loss of antioxidant compounds
Subsequent storage and heating
Inadequate manufacturing controls
These contaminants provide an additional reason to question routine dependence on heavily refined oils when minimally processed or more gently produced alternatives are readily available.
They do not establish that every refined seed oil contains a dangerous contaminant dose. Product quality and processing conditions vary widely, and modern mitigation strategies can substantially reduce formation.
But they strengthen the case for precaution: the concern is not limited to linoleic acid itself. It also includes what can happen to the oil before it reaches the consumer.
Replacing saturated fat with linoleic-acid-rich oil lowers LDL cholesterol.
What has not been proven is that increasing isolated omega-6 LA prevents cardiovascular events or extends life.
The Minnesota Coronary Experiment achieved substantial cholesterol reduction without a corresponding reduction in coronary events or total mortality.
A reanalysis of the Sydney Diet Heart Study reported higher cardiovascular and all-cause mortality among men assigned to a high-LA intervention. However, the study involved men with established coronary disease and used products from an era in which margarine composition and trans-fat exposure were not always documented clearly.
These historical studies do not cleanly prove that modern, nonhydrogenated liquid seed oils are harmful. They involved variable adherence, older food products, and multiple simultaneous dietary changes.
They also show that lowering LDL through omega-6 substitution cannot automatically be treated as proof of improved overall health.
A 2024 meta-analysis found no clear effect of increasing omega-6 or total polyunsaturated fat on cardiovascular events or mortality, with substantial heterogeneity and limited certainty.
Long story short, linoleic acid lowers LDL when it replaces saturated fat. But that doesn’t provide a clear benefit on cardiovascular events and mortality.
Large prospective studies generally associate higher circulating or adipose-tissue linoleic acid with lower cardiovascular risk, lower type 2 diabetes risk, and lower cardiovascular and all-cause mortality.
A pooled analysis of 30 prospective studies involving more than 68,000 participants found that higher LA biomarkers were associated with lower total cardiovascular risk, cardiovascular mortality, and ischemic stroke.
These findings provide meaningful reassurance. They make a large universal toxic effect unlikely.
However, they do not prove that refined seed oils are protective.
LA biomarkers reflect both intake and metabolism and cannot distinguish refined oils from nuts, seeds, dressings, or mixed foods. They also do not show whether an oil was repeatedly heated or whether tissue-specific oxidation occurred.
These associations should constrain the magnitude of concern. But do not fully resolve smaller cumulative effects associated with particular sources, preparation methods, or susceptible populations.
The effect of linoleic acid cannot be separated completely from the diet in which it appears.
Omega-3 intake, total calories, saturated-fat intake, fiber, antioxidant-rich foods, alcohol exposure, metabolic health, and the degree of food processing can all influence lipid metabolism and oxidative stress.
LA consumed through nuts and seeds in a minimally processed diet is not necessarily equivalent to the same amount supplied through repeatedly heated restaurant oil or packaged ultra-processed foods.
The replacement also matters.
Replacing butter with soybean oil generally lowers LDL (although this hasn’t been proven to improve outcomes).
Replacing extra-virgin olive oil with soybean oil removes olive polyphenols and increases the proportion of linoleic acid.
Replacing deep-fried food with minimally processed food changes calories, sodium, refined starch, fiber, food structure, and many other variables at once.
Removing seed oils does not automatically improve health if the overall diet remains dominated by excess calories and ultra-processed foods.
The practical question is not simply whether a food contains a seed oil. It is which oil is being used, how it was processed, how it was cooked, how much is consumed, and what replaces it.
A reasonable health decision can consider more than whether harm has been proven beyond doubt.
Relevant considerations include:
Whether the proposed mechanism is biologically credible
Whether potential harm could accumulate slowly
Whether current studies can detect it
Whether the exposure is nutritionally necessary
Whether safer or more stable alternatives are practical
Whether the substitution creates a different risk
Refined high-linoleic oils provide no unique nutritional benefit that cannot be obtained elsewhere. Essential linoleic-acid requirements are modest and can be met through a varied diet.
For many people, reducing routine exposure is straightforward:
Use extra-virgin olive oil or avocado oil as the primary household oil.
Use high-oleic oils when a neutral flavor or greater heat stability is needed.
Minimize commercial deep-fried foods and oils exposed to prolonged high-temperature use or multiple heating cycles.
Reduce ultra-processed foods in which refined oils are major calorie sources.
Obtain fats through whole foods and other minimally processed foods.
This is not a call to fear trace exposure or avoid every restaurant meal.
Personally, I don’t cook with soybean, corn, sunflower, or safflower oil at home.
At home, I use extra-virgin olive oil and avocado oil from reputable producers, along with butter and ghee.
This substitution costs me nothing; the thermal degradation chemistry is solid, and the refining contaminant data are real.
For me, this is a low-cost substitution with no meaningful practical downside. The thermal-degradation chemistry is established, and the refining-contaminant data are real. I do not need a decades-long outcome trial to justify preferring those alternatives in my own home.
But I also eat at restaurants, and I don’t ask about the fryer, and if I’m at a friend’s house and dinner was cooked, I eat dinner without interrogating my friends about which oil they used.
The concern, if valid, is more likely to involve routine exposure over years than occasional consumption.
Dietary linoleic acid changes human tissue composition, is more susceptible to oxidation than oleic acid, and can generate metabolites capable of altering mitochondrial and cellular function.
Human interventions confirm that changing LA intake can affect selected OXLAMs and oxidation markers.
What remains uncertain is whether those biochemical effects produce meaningful cumulative damage at typical intake levels.
Favorable LA biomarker associations make a large universal toxic effect unlikely. But they do not directly test decades of refined high-LA oil exposure, repeated heating, tissue-specific oxidation, or vulnerability in people with impaired redox control.
The evidence therefore supports neither panic nor complete reassurance. This is partly because the proper studies to answer this question haven’t been done, and probably won’t be done.
The strongest recommendation is to minimize repeatedly heated, rancid, and degraded oils. Limiting refined high-LA oils as a major daily fat source is also a reasonable precaution when they can be replaced with extra-virgin olive oil, high-oleic oils, and whole foods.
Like most things, occasional exposure is unlikely to determine anyone’s health. In my opinion, it depends on the dose, duration, condition of the oil, metabolic environment, and individual susceptibility. The problem is we don’t know what the individualized threshold is for each of those variables.
Routine exposure over decades is the unresolved question that most likely will never be answered.
If you want to dive deeper into understanding what you should do for your own metabolic health, then you should check out a brand new program we just released called The Metabolic Roadmap.
You’ll learn a clear, repeatable process for evaluating your metabolic health and deciding where to focus first.
You’ll learn the framework I use with private coaching clients—a process you can return to as you evaluate where you are, choose what to address, and adjust your approach over time.
→ Click here to learn more about The Metabolic Map
To better health, with better science.
– Andrew
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.