How Should You Choose a Cooking Oil? An Evidence-Based Appraisal of Coconut Oil, Seed Oils, and High-Heat Cooking

The Truth About Cooking Oils: Health Cannot Be Judged by a Single Label

The conclusion in one sentence

A cooking oil cannot be judged solely by labels such as “natural,” “plant-based,” “saturated fat,” or by its smoke point. A more reliable approach considers the food being replaced, the cooking method, and human clinical evidence together. For routine use, plant oils composed mainly of unsaturated fats remain preferable substitutes for sources high in saturated fat, such as butter. PMID: 40885400

Coconut oil affects blood lipids, but that does not mean cardiovascular events have been proven

In his original article, Professor Ching-Shung Lin opposed marketing coconut oil as a health panacea, citing its tendency to raise low-density lipoprotein cholesterol as an important reason. Compared with nontropical vegetable oils, pooled clinical-trial results support the conclusion that coconut oil raises low-density lipoprotein cholesterol. This is a major risk factor for atherosclerosis and demonstrates why the oil or fat being replaced cannot be ignored. PMID: 31928080

However, lipid risk factors and clinical events such as myocardial infarction or death represent different levels of evidence. Current data support the statement that coconut oil should not be regarded as heart-protective, but they are insufficient to assert directly that eating coconut oil inevitably causes cardiovascular disease or that it has no potentially offsetting effects whatsoever. Evidence that sustained reductions in saturated fat can lower combined cardiovascular events still supports a cautious approach to oils and fats high in saturated fat. PMID: 32827219

Results also vary with the oils, populations, and study durations being compared. A short-term trial in healthy adults found no significant difference between coconut oil and olive oil in their effects on low-density lipoprotein cholesterol, whereas butter raised it more than either oil. This finding does not overturn pooled comparisons with other vegetable oils, nor can short-term lipid changes be treated as proof of long-term safety. PMID: 29511019

Coconut oil, MCT, and dementia: the literature has advanced, but efficacy remains unestablished

When Professor Lin published the original article, he noted that there was no scientific evidence that coconut oil could slow dementia. The literature has since advanced: a meta-analysis has now synthesized a small number of studies and identified a favorable signal in cognitive scores. Therefore, “there is no research at all” no longer accurately describes the evidence available today. This reflects the evolution of science; it does not rewrite the professor’s historical position at the time. PMID: 39589946

A more rigorous placebo-controlled trial did not show that virgin coconut oil improved overall cognition, and heterogeneity prevented a systematic review from reaching a firm conclusion. The balanced interpretation remains that coconut oil is not an established treatment for dementia and cannot replace standard medical care. Favorable signals may justify further research, but they must not be prematurely translated into claims of health benefit. PMID: 37980665

Commercial MCT products generally contain particular medium-chain fats, whereas the fatty-acid composition of coconut oil is not equivalent to that of these products. Findings from MCT research therefore cannot be applied directly to coconut oil. The clinical role of MCT in Alzheimer’s disease remains very limited, and there is insufficient evidence to call it an effective treatment. PMID: 24413538

Seed oils and omega fatty acids: reject slogans from either direction

Current human evidence does not support categorically describing seed oils such as soybean oil and peanut oil as “highly toxic.” Reviews of human evidence on the major omega-6 fatty acids generally do not support the one-way narrative that a higher amount is necessarily less healthy. This is not an endorsement of unlimited consumption; it is a call to return to the context of the whole diet, the amount consumed, processing, and substitution. PMID: 41613930

A common claim is that omega-6 inevitably promotes inflammation. A systematic review of randomized trials found no evidence that increasing linoleic acid raises markers of chronic inflammation in healthy people, and subsequent pooled results likewise do not support the claim that omega-6-rich cooking oils necessarily cause systemic inflammation. Professor Lin’s objection to presenting that claim as a medical consensus is consistent with the direction of current evidence. PMID: 22889633

Conversely, plant-derived ALA cannot be equated directly with EPA and DHA in fish oil merely on the basis of “total omega-3 content,” because adults have limited capacity to convert ALA into long-chain omega-3 fatty acids. This is also why products such as sacha inchi oil cannot claim the same physiological effects as fish oil based only on the total amount printed on their packaging. PMID: 17622276

The key question about saturated fat is what replaces it

Dietary research asks not only what people eat less of, but also what they eat instead. Pooled evidence from randomized trials supports the conclusion that reducing saturated fat over the long term can lower combined cardiovascular events. Replacement with polyunsaturated fat is particularly relevant for people at high cardiovascular risk. This supports Professor Lin’s general position that lard, butter, and coconut oil should not serve as the principal everyday fats, but it cannot be extended to imply an identical magnitude of benefit for every person, every oil, and every mortality outcome. PMID: 32428300

Large-scale data associating the replacement of butter with plant oils with a lower risk of death point in the same direction. These findings are observational, however, and cannot by themselves prove that any particular plant oil necessarily extends life. The most defensible statement remains that the substitution may be beneficial, not that any single oil deserves an unconditional endorsement. PMID: 40048719

Ghee likewise should not be treated as a therapeutic product simply because it has a traditional or natural image. A randomized crossover trial in humans found that a diet centered on ghee produced less favorable values for some cardiometabolic risk markers than a diet centered on olive oil. As for claims that ghee repairs the gut, treats leaky gut, or reduces systemic inflammation, ECU currently has no human clinical evidence sufficient to establish such effects. PMID: 34794522

High-heat cooking: smoke point is not the only answer

In general, oils with a higher degree of saturation have greater oxidative stability, while polyunsaturated fats degrade more readily when heated. This is an important principle of food chemistry. Yet “more stable chemically” does not translate directly into “healthier when consumed over the long term,” because cooking stability and cardiovascular outcomes in humans answer different questions. PMID: 39767128

Using smoke point alone to determine whether an oil is suitable for frying is also an oversimplification. The literature has since advanced: although smoke point is related to thermal deterioration, it is not a sufficiently reliable stand-alone criterion for choosing an oil. Fatty-acid composition, antioxidant constituents, heating duration, temperature, and repeated use all influence degradation. This is a science-based refinement of an older, simplified rule. PMID: 41376035

Extra-virgin olive oil has relatively good oxidative stability, and studies suggest that it may undergo less degradation during heating and frying. Flavor, price, and suitability for a particular dish are practical considerations, not questions of clinical efficacy. Its higher price does not prove that it is necessarily healthier, and assumptions about its smoke point do not justify declaring it categorically unsuitable for high-heat cooking. PMID: 41744988

Heating plant oils to high temperatures can indeed generate aldehydes with potential toxicity. Detecting a chemical, however, does not prove that ordinary consumption causes cancer, much less establish that switching to lard or butter would improve health. Professor Lin’s insistence on distinguishing chemical formation, exposure level, and clinical outcomes in actual people defines a correct and necessary boundary for the evidence. PMID: 40698373

Refining contaminants, palm oil, and risk communication

High-temperature refining of plant oils can produce process contaminants such as glycidyl esters. Palm oil and infant formula containing vegetable oils may raise concerns about greater exposure, making process control and monitoring necessary. The carcinogenic concern surrounding these compounds comes mainly from genotoxic mechanisms and animal data; it cannot be restated as proof that currently marketed formula causes cancer in human infants. PMID: 39335805

Palm oil has a relatively high proportion of saturated fat, and its effects on blood lipids compared with unsaturated oils deserve attention. Overall cardiovascular risk, however, still depends on dietary pattern, the replacement food, and clinical endpoints. In the original article, Professor Lin separated animal evidence of carcinogenicity from human evidence and withheld a definitive conclusion on the health effects of palm oil. That discipline—avoiding the portrayal of a risk signal as certain harm—remains worth following. PMID: 39497694

Practical principles for choosing an oil

When choosing an everyday oil, first ask whether it is replacing a source high in saturated fat, such as butter or ghee, or another unsaturated plant oil. Consider whether it will be used in a cold dressing, a brief stir-fry, or prolonged and repeated high-heat cooking. Finally, determine whether the product claim rests on composition, animal research, blood markers, or actual disease outcomes in humans. Once these levels are separated, much of the rhetoric about “miracle oils” and “toxic oils” loses its persuasive force. PMID: 39767128

Canola oil comes from specific low-erucic-acid cultivars; not every product described without qualification as rapeseed oil can be treated as the same cooking oil. Current data suggest no clear short-term harm in certain contexts and possible improvements in some metabolic markers. Long-term clinical outcomes are lacking, however, so the evidence does not justify claiming that every use is necessarily healthy. PMID: 42016563

Ultimately, there is no need to search obsessively for the single “healthiest” oil. Using plant oils composed mainly of unsaturated fats as the everyday foundation, avoiding repeated high-temperature use, and placing the whole diet, disease risk, and intended use ahead of any single nutrition label is the approach most consistent with current evidence. PMID: 41397264