Saturated fat: the answer depends on what replaces it

Cutting saturated fat lowered heart attacks, strokes and related events by 17% in the trials that tested it.¹ That headline hides the two things that decide whether it means anything to you. What took the fat's place on the plate. And how likely you already were to have a heart attack.²

What the trials actually asked

The anchor is a review of trials in which adults were assigned either to cut saturated fat or to carry on eating as before, for at least two years.¹ Fifteen trials and 56,675 people qualified. Some cooked and delivered every meal. Some only handed out advice.¹

Combined heart events fell 17%.¹ The range around that is wide: the true reduction could be as large as 30% or as small as 2%.¹ Deaths barely moved at all, for any cause or for heart disease specifically.¹

Put in people rather than percentages, in trials of adults without existing heart disease, 56 had to cut saturated fat for about four years for one of them to avoid an event.¹

Cutting saturated fat for about four years, adults without heart disease one event avoided no difference

The dose did the work, not the instruction

One finding inside that review matters more than the headline. The trials that cut saturated fat the hardest, and so dropped blood cholesterol the furthest, produced the biggest reductions in events — and that pattern explained most of the disagreement between trials.¹

The trials differed widely in how much saturated fat they actually removed, from handing over every meal to giving advice. The ones that removed more saw bigger drops in events.

Replaced by what: three answers that do not agree

The Cochrane review's own split came out undecided. Swapping saturated fat for polyunsaturated fat and swapping it for carbohydrate produced no clear difference, and there was too little data on monounsaturated fat or protein to judge either.¹

A 2026 review of 17 trials and 66,337 people found the one place the replacement did show up. When polyunsaturated fat took the saturated fat's place, nonfatal heart attacks fell about a quarter — a reduction that could be as large as 42% or as small as 1%.² Monounsaturated fat and protein were thinly tested there too.²

A 2017 review of long-running observational studies goes furthest. Across two large cohorts of nurses and health professionals, a diet high in saturated and trans fat came with 8-13% higher mortality, and replacing saturated fat with carbohydrate, polyunsaturated or monounsaturated fat all tracked lower mortality — polyunsaturated fat the more effective of the two fats, 19% against 11%.⁴ For heart deaths only polyunsaturated fat and fish oil helped, by 28% for each 5% of daily calories swapped.⁴ Swapping in whole grains lowered coronary events. Swapping in sugar and starch raised them.⁴

These are people who chose their own diets, not people assigned to one. That review's own conclusion is narrower than its numbers: replacing saturated fat with carbohydrate will not lower heart events, while replacing it with polyunsaturated fat, monounsaturated fat or good-quality carbohydrate will.⁴

Why swapping beats adding

The cleanest mechanism comes from trials that fed people everything they ate. Twenty-four such trials in 1,011 healthy adults, all matched for calories, fiber and total fat, compared diets by their ratio of polyunsaturated to saturated fat.³ The higher-ratio diets lowered LDL cholesterol by 9.83 mg/dL — about ten points on a cholesterol test.³

Then the result that settles the argument at the level of blood. Where the two diets differed in saturated fat by at least 2% of calories, LDL fell 15.72 mg/dL. Where saturated fat was held equal and only the polyunsaturated fat changed, the drop shrank to 3.45 mg/dL and stopped being distinguishable from chance.³

LDL cholesterol change on higher vs. lower polyunsaturated-to-saturated diets, mg/dL0 = no effect

Diets differed in saturated fat-15.72 (-20.51 to -10.92)

Saturated fat held equal-3.45 (-7.88 to 0.98)

Pouring more sunflower oil on top of the same amount of butter does almost nothing. Using it instead of the butter does nearly all of the work.

Who the benefit was actually for

The 2026 review is the one that sorts people by their starting risk, and in it that split changes more than the replacement question does. The Cochrane review reached a different view on this point: it found that the reduction in events did not significantly change with baseline cardiovascular risk.¹

It set thresholds in advance for what counts as worth having: 5 fewer deaths per 1,000 people over five years, or 10 fewer nonfatal events.² For adults at low heart risk, cutting saturated fat came in below both.² For adults at high risk, it came in above.² The review's verdict for low-risk adults is blunt — little or no benefit over five years.²

A survey of 21 previous analyses lands in the same shape. Across the trials, combined heart events fell 21%, judged moderately certain, while deaths from any cause, from heart disease and from cancer showed nothing.⁵ Across the observational studies, people eating more saturated fat had a 10% higher rate of death from coronary heart disease.⁵

A percentage cut off a big risk is worth more than the same cut off a small one. That is most of why the guidance and the trials appear to disagree.

Where saturated fat did not behave as expected

Two feeding trials complicate the story without overturning it.

In the first, 164 people who had already lost 10-14% of their body weight spent 20 weeks on diets that differed threefold in both carbohydrate and saturated fat, running from 20% carbohydrate with 21% saturated fat up to 60% carbohydrate with 7% saturated fat.⁷ LDL cholesterol, LDL particle number and inflammatory markers came out the same on all three.⁷ The lowest-carbohydrate arm lowered lipoprotein(a), a heart-risk particle, by 14.7%.⁷ A diet one-fifth saturated fat did not raise LDL when carbohydrate fell alongside it.

The second went the other way. Seventeen healthy young women ate a ketogenic diet at 4% carbohydrate and 77% fat, and a control diet at 44% carbohydrate and 33% fat, four weeks on each.⁶ LDL cholesterol rose in every single woman, by 1.82 mmol/L.⁶

Same nutrient, opposite directions. The two trials differed in who ate the diet and in how much else changed alongside the saturated fat.

What remains unknown

Monounsaturated fat is the standing gap. Both the Cochrane review and the 2026 review report too little data on replacing saturated fat with monounsaturated fat or protein to say anything, which leaves the olive-oil version of the question effectively untested.^1,2

The second gap is that none of this is about food. Every trial manipulates a class of nutrient, not a product. Nothing here tested coconut oil, or butter against a named spread, so a claim about a jar is an extrapolation from a claim about a percentage of calories.

The third is the disagreement itself. One analysis saw no difference between polyunsaturated fat and carbohydrate as the replacement; the other two saw polyunsaturated fat doing more.^1,2,4 The trials varied a great deal in how much saturated fat they removed and in what they put in its place, which is why the calls for new trials have not stopped.²