Are seed oils bad for you?

The larger body of evidence does not show seed oils as harmful to the heart, but it does not settle every question either. One recovered trial from the 1970s found that swapping saturated fat for corn oil lowered cholesterol without lowering deaths1. Pooled studies of tens of thousands of people point the other way, linking more of the main seed-oil fat with fewer heart and stroke events2.

This article sets out both sides of the nutrition research exactly: what each study measured, and what it cannot show. It describes research; it does not tell anyone to use or avoid an oil. For questions about your own heart or cholesterol, a doctor is the right person to ask.

First, the words

Seed oils are oils pressed from seeds, such as sunflower, corn, soybean and rapeseed oil. Their main fat is linoleic acid, an omega-6 fat. Omega-6 is a family of polyunsaturated fats, the kind that stays liquid at room temperature.

Saturated fat is the kind found mostly in butter, fatty meat and cheese. Most of the seed-oil debate is really a question of swaps: what happens when linoleic acid takes the place of saturated fat.

Two kinds of study come up again and again:

• A randomised trial lets chance decide who gets which diet. It can show cause, if it is large, long and well run.
• A cohort study follows people for years and sees who gets ill. It shows associations: two things moving together, which is not proof that one causes the other.

A hazard ratio compares how often an event happens in one group against another. A value of 1 means no difference; 0.78 means the event happened about 22% less often in the higher group.

In plain terms: the question is less "is this oil toxic" and more "what happens to heart outcomes when this fat replaces that one".

The case for worry: the Minnesota trial

The worry leans heavily on one old trial. The Minnesota Coronary Experiment ran from 1968 to 1973 in one nursing home and six state mental hospitals1. It randomised 9,423 women and men aged 20 to 97.

One group had saturated fat replaced by corn oil and corn-oil margarine. The control diet was high in saturated fat from animal fats, common margarines and shortenings1.

Its full results were never published at the time. In 2016, researchers recovered the raw data and a 1981 thesis that held the original survival analyses1.

What they found:

• Cholesterol fell far more on corn oil: a mean change of −13.8% against −1.0% in the control group1.
• Deaths did not fall. There was no mortality benefit in the full group or in any planned subgroup1.
• A bigger fall in cholesterol went with more deaths. Each 30 mg/dL drop was associated with a 22% higher hazard of death (hazard ratio 1.22, 95% confidence interval 1.14 to 1.32)1. A confidence interval is the range the true value most likely sits in.
• Five similar trials combined, 10,808 people, showed no benefit on heart-disease deaths (risk ratio 1.13, 0.83 to 1.54) or deaths from any cause (1.07, 0.90 to 1.27)1.

This is the strongest data point on the worried side. It is also narrower than it looks. The authors write that the cholesterol-and-death link is "observational in nature", even though it sits inside a randomised trial1. Only 2,355 people stayed on the study diets for a year or more. Everyone lived in an institution. The intervention diet held almost twice the linoleic acid of the average American diet, and its margarine likely contained some trans fat. The authors also say the results "should not be generalized to nuts or other unprocessed foods containing linoleic acid"1.

In plain terms: in this trial, the oil swap lowered cholesterol without saving lives, over a short stay in hospital settings.

The other side: the cohort evidence

The largest pooled analysis took a different approach. Instead of asking people what they ate, it measured linoleic acid in their blood or body fat, a marker of long-term intake2.

It combined individual data from 30 studies in 13 countries: 68,659 adults free of cardiovascular disease at the start, with median follow-up ranging from 2.5 to 31.9 years across studies2. Over that time, 15,198 cardiovascular events occurred.

Comparing high with low linoleic acid levels (the gap between the bottom and top fifth), higher levels went with2:

• Total cardiovascular disease: hazard ratio 0.93 (0.88 to 0.99), about 7% fewer events.
• Cardiovascular death: 0.78 (0.70 to 0.85), about 22% fewer.
• Ischaemic stroke (a stroke from a blocked artery): 0.88 (0.79 to 0.98).
• Coronary heart disease: 0.94 (0.88 to 1.00), which does not reach statistical significance.

Arachidonic acid, a fat the body makes from linoleic acid, was not linked to more cardiovascular events either2.

This is association, not cause. The authors note that residual confounding cannot be excluded: people with more linoleic acid may differ in ways the models did not capture2. Fatty acids were measured once, and most participants were of European descent. A food company gave a restricted grant that partly funded the analysis; the authors state it had no role in the design, analysis or writing2.

A 2025 study pooled three further large cohorts, with 172,891 people, and updated the meta-analyses. Linoleic acid was associated with fewer strokes (hazard ratio 0.89 per standard deviation, 0.83 to 0.95), and the pooled literature suggested inverse associations with both heart disease and stroke4. In its largest cohort, though, several omega-6 fats the body makes from linoleic acid were associated with more coronary heart disease. The authors called for more randomised trials.

In plain terms: across very large groups, people with more of the main seed-oil fat in their bodies had fewer heart and stroke events, but that does not prove the oil caused it.

What the trials of saturated fat show

A Cochrane review is a systematic review run to strict, published rules. The 2020 review on cutting saturated fat pooled 15 randomised trials, about 59,000 people, each lasting at least two years3.

Cutting saturated fat went with 21% fewer combined cardiovascular events (risk ratio 0.79, 0.66 to 0.93; moderate-quality evidence). It had little or no effect on deaths from any cause (0.96) or cardiovascular deaths (0.95)3. The authors concluded that replacing saturated fat with polyunsaturated fat or carbohydrate "appear to be useful strategies".

Two limits matter here. The review tested cutting saturated fat, not seed oils on their own. And we read its abstract only, so we cannot say which trials used which oils.

In plain terms: the trials that cut saturated fat for years saw fewer heart events on average, though not fewer deaths.

What about inflammation?

A common claim online is that seed oils cause inflammation. The newest trial we found tested exactly that.

In a 2026 crossover trial, 10 generally healthy adults with overweight or obesity ate study foods delivering 30 g of soybean oil or 30 g of palm oil a day, each for 4 weeks, in random order5. Most inflammation markers and oxidised LDL did not change. Interleukin-6, one inflammation marker, showed a non-significant trend downward on soybean oil.

It is a pilot study. The authors say its small size makes it "inappropriate to generalize" to the wider adult population, and it was funded in part by a soybean industry board5. It is one small data point, not a verdict.

What the studies found

The nuance

Both sides are easy to misread.

• "Minnesota proves seed oils kill." It does not. The death signal was an association inside the trial, the follow-up was short, and the setting was institutions1.
• "The cohorts prove seed oils protect." They do not. They show that people with more linoleic acid had fewer events, which could reflect other habits2.
• An oil is not a diet. Seed oils often arrive in fried food, baked goods and snacks. A study of the oil itself does not tell you about the foods it comes in.

Fat is also calorie-dense, at 9 kcal per gram, whatever the source. If you track intake, see how much fat a day and what macros are for how fat fits into a daily total.

Limits worth knowing

• The key trial is old and short. Minnesota ran in the late 1960s, in institutions; only about a quarter of participants were on the diets for a year or more1.
• The cohort data are observational. Residual confounding cannot be ruled out, and fatty acids were measured once2.
• Abstract only for two papers. We read the Cochrane review and the 2025 cohort study as abstracts.
• The review tested saturated fat, not oils. It cannot isolate seed oils3.
• The inflammation trial was tiny. Ten people, four weeks per diet, partly industry funded5.
• Funding. The 2019 pooled analysis was partly supported by a restricted food-company grant2.
• What was not measured. None of these studies tested frying oil reused many times, or foods rather than oils.

Latest research

The newest study we opened is the 2026 soybean-oil crossover trial: 30 g a day for 4 weeks did not move most inflammation markers in 10 adults5. The newest large dataset is the 2025 cohort analysis, which again linked linoleic acid with fewer strokes4. Neither is the long, large randomised trial that would settle the question. When one appears, this page will be updated and its date will change.

The short version

• The worried case rests mostly on the recovered Minnesota trial: cholesterol fell on corn oil, deaths did not.
• Large cohort studies link more linoleic acid with fewer heart and stroke events; that is association, not cause.
• Trials that cut saturated fat, often replacing it with polyunsaturated fat, saw fewer combined heart events on average.
• A small 2026 trial found no change in most inflammation markers on soybean oil.
• The evidence does not show seed oils as harmful, and it does not settle every question.

Educational overview only — not medical advice.