Tempeh and tofu: a small real cholesterol effect, and nobody has compared the two

The cholesterol effect is real and it is small. Across 43 of the 46 controlled trials the FDA itself assembled, in adults whose starting LDL ran from 110 to 201 mg/dL, a median 25 g of soy protein a day over a median six weeks lowered LDL cholesterol by 4.76 mg/dL, with a confidence interval from 2.80 to 6.71 mg/dL of reduction, which works out to roughly 3 to 4 percent. Heterogeneity was high, at 55 percent for LDL and 74 percent for total cholesterol, only a minority of the individual trials were significant on their own, and the trials fed soy protein rather than a block of tofu or a slab of tempeh. Two regulators looked at this and disagreed. United States law still carries the model wording that 25 grams of soy protein a day may reduce the risk of heart disease, and the FDA proposed revoking that claim in 2017 without settling the question since. The EU register refused the same cholesterol claim, refused the wider promises about vascular benefit and artery health, and across 26 soy claims assessed it authorized none. Refusal means the evidence fell short at the time of assessment, not that the effect was disproved. The hormone worry has been tested properly and it did not hold up. In 40 randomized trials covering 3285 postmenopausal women, a median 75 mg of soy isoflavones a day for a median 24 weeks left every measure of estrogenicity unmoved: endometrial thickness, vaginal maturation index, FSH and estradiol all sat on zero, at high to moderate certainty. Those women were postmenopausal, and the same question in men and in children was not what these trials asked.

How much is actually in it

• Tempeh carries 20.7 g protein and 166 calories per 3.5 oz (100 g), against 8.1 g protein and 73 calories for steamed tofu. — CoFID 2021 reference entries Tempeh (code 13-118) and Tofu, soya bean, steamed (code 13-570) (n = 2 records) Evidence E source
  Tempeh per 3.5 oz (100 g): protein 20.7 g, fat 6.4 g, carbohydrate 6.4 g, NSP fiber 4.3 g, 166 calories; tofu steamed per 3.5 oz (100 g): protein 8.1 g, fat 4.2 g, carbohydrate 0.7 g, 73 calories, water 85.0 g

What a real portion looks like

• The trials behind the cholesterol claim used a median 25 g of isolated soy protein a day, a study dose rather than a serving of tofu or tempeh. — 43 controlled trials in adults identified by the FDA for its review of the soy protein claim (n = 43 trials) Evidence A source
  Median dose 25 g soy protein per day, median follow-up 6 weeks

What your body absorbs

• Only 31 percent of 498 adults reliably turned soy daidzein into equol, and 43 percent never did, on the same soy intake. — 498 community-dwelling adults, annual health checkups 2016 to 2019, urinary equol and daidzein measured each year (n = 498) Evidence C source
  30.9 percent stable producers, 26.1 percent unstable, 43.0 percent non-producers; stable producers had higher gut bacterial diversity despite comparable daidzein intake
• Tempeh holds 120 mg calcium, 3.60 mg iron, 0.67 mg copper and 1.40 mg manganese per 3.5 oz (100 g). — CoFID 2021 reference entry Tempeh, code 13-118 (n = 1 record) Evidence E source
  Calcium 120 mg, iron 3.60 mg, copper 0.67 mg, magnesium 70 mg, manganese 1.40 mg, phosphorus 200 mg per 3.5 oz (100 g)

What cooking and storage change

• Fermented soy, the family tempeh belongs to, cut waist circumference by 0.59 in (1.50 cm) and fasting glucose by 6.39 mg/dL across 15 randomized trials. — participants of 15 randomized controlled trials of fermented soy products, searched to March 2024 (n = 15 trials) Evidence A source
  BMI -0.14 kg/m2 (95 percent CI -0.28 to -0.01), waist -0.59 in (-1.2 to -0.03; metric: -1.50 cm, -2.94 to -0.07), visceral fat -692.17 mm2 (-1011.58 to -372.77), fasting glucose -6.39 mg/dL (-10.38 to -2.40), total cholesterol -5.0 mg/dL (-6.60 to -3.39)
• Fermenting soy can add vitamin K2: human studies on natto showed improved vitamin K2 status. — human interventional and observational studies on fermented foods, literature 1970 to 2024, EFSA and COST Action PIMENTO protocols Evidence C source
  Vitamin K2 status effectively improved in several studies on natto; microorganisms raise vitamin K2, B2, B9 and B12 levels in fermented foods

What it does to your health

• Soy protein at a median 25 g a day for a median 6 weeks lowered LDL cholesterol by 4.76 mg/dL, about 3 to 4 percent. — adults with baseline LDL cholesterol 110 to 201 mg/dL, 43 of the 46 controlled trials the FDA itself identified (n = 43 trials) Evidence A source
  LDL cholesterol -4.76 mg/dL (95 percent CI -6.71 to -2.80), total cholesterol -6.41 mg/dL (95 percent CI -9.30 to -3.52); no dose-response and no publication bias
• Soy protein supplements did not build lean body mass in 261 young adults doing resistance training across 12 trials. — trained and untrained adults aged 18 to 30 engaged in resistance exercise training, 12 studies (n = 261) Evidence A source
  No significant effect of whey or soy protein on lean body mass; whey improved bench press by 8.87 (95 percent CI 5.95 to 11.79) and squat by 9.60 (5.61 to 13.60), soy did not

Safety, limits and interactions

• The FDA label for levothyroxine says soybean flour may bind the drug and cut its absorption from the gut. Tempeh and tofu are soy foods but the label does not name them. — People taking levothyroxine sodium tablets (n = -) Evidence E source
  Direction only: soybean flour may bind levothyroxine and reduce its absorption; no size, no amount of soy, no timing in this section
• The FDA label for transdermal selegiline (EMSAM) at 9 mg/24 h or more puts most soybean products on the avoid list, naming soy sauce and tofu. Tempeh is a soybean product but is not named. — Adults using EMSAM (selegiline transdermal system) at 9 mg/24 h or 12 mg/24 h (n = -) Evidence E source
  Avoid list in label Table 5, miscellaneous class
• Soy milk dropped one stable warfarin patient below his therapeutic INR, and the value came back within two weeks of stopping it. — one 70-year-old man stable on warfarin therapy who started drinking soy milk (n = 1) Evidence D source
  Subtherapeutic INR while drinking soy milk; INR returned to therapeutic range within 2 weeks of stopping; causality judged possible to probable

What people believe that the data does not support

• United States law still carries a model wording tying 25 grams of soy protein a day to lower heart disease risk. — United States food labeling, 21 CFR 101.82, model health claim text Evidence E source
  Authorized model claim naming 25 grams of soy protein a day as part of a diet low in saturated fat and cholesterol
• Europe refused the same cholesterol claim the United States allows: the soy protein and cholesterol wording is non-authorized in the EU register. — EU Register on nutrition and health claims, snapshot 2026-09-21, claim POL-HC-6237 Evidence E source
  Status Non-authorized; across the register 26 soy claims were assessed and none was authorized
• The broader vascular promises made for soy were also refused in Europe, including that soy keeps the arteries healthy. — EU Register on nutrition and health claims, snapshot 2026-09-21, claim POL-HC-7954 Evidence E source
  Status Non-authorized for vascular benefit, artery health and antioxidant protection wording

Who this works differently for

• Soy isoflavones at a median 75 mg a day for 24 weeks moved none of the four estrogen markers in 3285 postmenopausal women. — postmenopausal women, 40 randomized trials of at least 3 months, 52 trial comparisons (n = 3285) Evidence A source
  Endometrial thickness -0.22 mm (95 percent CI -0.45 to 0.01), vaginal maturation index 2.31 (-2.14 to 6.75), FSH -0.02 IU/L (-2.39 to 2.35), estradiol 1.61 pmol/L (-1.17 to 4.38); certainty high to moderate
• In 522 people with chronic kidney disease, swapping animal protein for soy protein lowered total cholesterol, LDL and proteinuria across 18 trials. — adults with chronic kidney disease, 18 randomized controlled trials published up to March 2024 (n = 522) Evidence A source
  Total cholesterol -20.55 mg/dL, LDL -8.26 mg/dL, proteinuria -140.53 mg/day; no change in creatinine, triglycerides, calcium, C-reactive protein, HDL, uric acid or phosphorus

What is still unknown

• The same review states there are still no well-controlled large-scale human studies validating fermented foods as a way to improve vitamin status. — human studies on fermented foods and vitamin status, literature 1970 to 2024 Evidence C source
  Evidence for B1, B2, B3, B5, B6 and B12 limited or inconsistent; no human evidence at all for other vitamins

The bottom line

On the shelf the two foods look interchangeable and the labels are not. Per 100 g, tempeh gives 20.7 g protein and 166 calories, steamed tofu 8.1 g protein and 73 calories, and most of that gap is water, because tofu is 85 percent water by weight. Tempeh also carries 120 mg calcium, 3.60 mg iron, 0.67 mg copper, 70 mg magnesium and 1.40 mg manganese per 3.5 oz (100 g). The iron is plant iron, and how much of it you actually take up was not measured in anything we found. The trials are no help with portion size either. They dosed isolated soy protein, and no source we have translates 0.88 oz (25 g) of it into a piece of tofu or tempeh you could put on a plate. Fermentation is where tempeh might differ, and the evidence treats it as a family rather than a food. Fifteen randomized trials of fermented soy products, natto and miso and douchi and tempeh pooled together with tempeh never separated out, found waist circumference down 0.59 in (1.50 cm) (95 percent CI 0.07 to 2.94), fasting glucose down 6.39 mg/dL (2.40 to 10.38), visceral fat down 692 mm2 and total cholesterol down 5.0 mg/dL. Those are small absolute numbers and nothing else in the panel moved. The vitamin K2 story that follows fermented soy around belongs to natto, which is made with Bacillus subtilis, while tempeh is made with a Rhizopus mold, and no study has checked whether the K2 finding carries across. Reviewers of the whole fermented food literature say there are still no large well-controlled human trials on vitamin status. Your own gut decides part of this. Among 498 Japanese adults measured yearly, 30.9 percent reliably converted soy daidzein into equol, 26.1 percent did so on and off and 43.0 percent never did, on comparable daidzein intake, with the steady converters carrying more diverse gut bacteria. That is observational, so which way the arrow points is unsettled, and the numbers come from one Japanese cohort. Two other groups get a different answer than the average reader. In 18 trials totaling 522 people with chronic kidney disease, replacing animal protein with soy protein lowered total cholesterol by 20.55 mg/dL, LDL by 8.26 mg/dL and proteinuria by 140.53 mg a day, with creatinine and phosphorus unchanged, on a small enough body of work that the authors flag the precision. In 12 trials of 261 adults aged 18 to 30 doing resistance training, soy protein powder produced no gain in lean body mass, and whey raised bench press and squat performance where soy did not. The safety column here holds exactly one report, and that is what our search found rather than a clearance for soy alongside medicines in general. We have no card on soy with thyroid medicine such as levothyroxine, or on tyramine in fermented soy such as tempeh for people taking MAOI antidepressants, so if you take either, ask your doctor or pharmacist about soy. A 70-year-old man stable on warfarin fell below his therapeutic INR after he started drinking soy milk, and the value came back within two weeks of stopping, which the authors rate as possible to probable causation rather than proven. That is one person and a different soy food, and if you take warfarin it is a reason to tell the person managing your dose that your soy intake changed. We have no card on soy allergy, and that is a gap in our sources, not a sign that soy is low risk for people who react to it: tempeh and tofu are both soy. We also found nothing comparing tempeh against tofu in a single trial, which means the question this page is named for has never been asked.

Sources

1. tftm-daily-r1-14 · food composition dataset
   CoFID 2021, Tempeh (13-118) per 100 g: Protein 20.7 g, Fat 6.4 g, Carbohydrate 6.4 g, NSP 4.3 g, Energy 166 kcal, Water 61.0 g. Tofu, soya bean, steamed (13-570) per 100 g: Protein 8.1 g, Fat 4.2 g, Carbohydrate 0.7 g, Energy 73 kcal, Water 85.0 g
   CoFID 2021 (Public Health England), codes 13-118 and 13-570 · checked 2026-09-25
2. tftm-daily-r1-02 · meta-analysis of RCTs
   Of the 46 trials identified by the FDA, 43 provided data for meta-analyses. Of these, 41 provided data for LDL cholesterol, and all 43 provided data for TC. Soy protein at a median dose of 25 g/d during a median follow-up of 6 wk decreased LDL cholesterol by 4.76 mg/dL
   J Nutr, 2019 · checked 2026-09-25
3. tftm-daily-r1-09 · prospective cohort
   Of the 498 participants, 30.9% were stable producers, 26.1% unstable producers, and 43.0% non-producers. Stable producers showed significantly higher gut bacterial diversity than unstable producers and non-producers and a distinct microbiota composition, despite comparable daidzein intake across groups.
   BMC Microbiol, 2026 · checked 2026-09-25
4. tftm-daily-r1-15 · food composition dataset
   CoFID 2021, Tempeh (13-118), Inorganics per 100 g: Calcium 120 mg, Copper 0.67 mg, Iron 3.60 mg, Magnesium 70 mg, Manganese 1.40 mg, Phosphorus 200 mg
   CoFID 2021 (Public Health England), code 13-118 · checked 2026-09-25
5. tftm-daily-r1-07 · meta-analysis of RCTs
   Results demonstrated significant beneficial effects of fermented soy on body mass index (WMD = -0.14 kg m-2, 95% CI: -0.28, -0.01, P = 0.039), waist circumference (WMD = -1.50 cm, 95% CI: -2.94, -0.07, P = 0.04), visceral fat (WMD = -692.17 mm2, 95% CI: -1011.58, -372.77, P < 0.001), fasting plasma glucose (WMD = -6.39 mg dL-1, 95% CI: -10.38, -2.40, P = 0.002), and total cholesterol (WMD = -5.0 mg dL-1, 95% CI: -6.60, -3.39, P < 0.001) compared with controls.
   Food Funct, 2025 · checked 2026-09-25
6. tftm-daily-r1-11 · systematic narrative review
   Findings confirm that certain microorganisms, including Bacillus subtilis, Propionibacterium freudenreichii, and some lactic acid bacteria, can increase the levels of vitamins K2, B2, B9, and B12 in FF. ... while vitamin K2 status was effectively improved in several studies on natto (fermented soy) and in one study on Jarlsberg cheese.
   Front Nutr, 2025 · checked 2026-09-25
7. tftm-daily-r1-01 · meta-analysis of RCTs
   Soy protein at a median dose of 25 g/d during a median follow-up of 6 wk decreased LDL cholesterol by 4.76 mg/dL (95% CI: -6.71, -2.80 mg/dL, P < 0.0001; I2 = 55%, P < 0.0001) and decreased TC by 6.41 mg/dL (95% CI: -9.30, -3.52 mg/dL, P < 0.0001; I2 = 74%, P < 0.0001) compared with non-soy protein controls.
   J Nutr, 2019 · checked 2026-09-25
8. tftm-daily-r1-13 · meta-analysis of RCTs
   A total of 12 studies with 261 participants were included after full-text screening with the main reason for exclusion being wrong population. The analysis found no significant effect of either whey or soy protein supplementation on LBM.
   J Diet Suppl, 2026 · checked 2026-09-25
9. sf-soylev-r5-6 · agency advice
   Soybean flour, cottonseed meal, walnuts, and dietary fiber may bind and decrease the absorption of levothyroxine sodium from the gastrointestinal tract.
   Levothyroxine sodium tablets, FDA-approved label, section 7.9, effective 2026-03-24 · checked 2026-10-08
10. sf-tymaoi-8 · agency advice
    Concentrated yeast extract (e.g., Marmite), sauerkraut, most soybean products (including soy sauce and tofu)
    EMSAM (selegiline transdermal system), FDA-approved label, sections 2.3 and 7.2 (Table 5) · checked 2026-10-08
11. tftm-daily-r1-10 · case report
    A 70-year-old white man who was stable on warfarin therapy developed subtherapeutic INR values after ingesting soy protein in the form of soy milk. ... INR values returned to therapeutic concentrations within 2 weeks after discontinuation of the soy milk.
    Ann Pharmacother, 2002 · checked 2026-09-25
12. tftm-daily-r1-03 · regulation
    25 grams of soy protein a day, as part of a diet low in saturated fat and cholesterol, may reduce the risk of heart disease. A serving of [name of food] supplies ____ grams of soy protein.
    21 CFR 101.82, Health claims: Soy protein and risk of coronary heart disease (CHD) · checked 2026-09-25
13. tftm-daily-r1-04 · regulatory decision
    POL-HC-6237 Isolated soy protein Protein-rich soybean component has been shown to lower/reduce blood cholesterol; blood cholesterol lowering may reduce the risk of (coronary) heart disease. -/- Non-authorised
    EU Register on nutrition and health claims, claim POL-HC-6237, snapshot 2026-09-21 · checked 2026-09-25
14. tftm-daily-r1-05 · regulatory decision
    POL-HC-7954 Soy -soy offers specific vascular benefits; '-soy helps keep the arteries healthy and helps keep a healthy heart; -soy helps protect the body tissues and cells from oxidation oxidative damage. Non-authorised
    EU Register on nutrition and health claims, claim POL-HC-7954, snapshot 2026-09-21 · checked 2026-09-25
15. tftm-daily-r1-06 · meta-analysis of RCTs
    We included 40 trials (52 trial comparisons, n = 3285) assessing the effect of a median reported dose of 75 mg/d of soy isoflavones in substitution for non-isoflavone controls over a median of 24 wk. Soy isoflavones had no statistically significant effect on any measure of estrogenicity; ET [mean difference, -0.22 mm (95% confidence interval, -0.45, 0.01 mm), PMD = 0.059], VMI [2.31 (-2.14, 6.75), PMD = 0.310], FSH [-0.02 IU/L (-2.39, 2.35 IU/L), PMD = 0.987], and estradiol [1.61 pmol/L (-1.17, 4.38 pmol/L), PMD = 0.256].
    Adv Nutr, 2025 · checked 2026-09-25
16. tftm-daily-r1-08 · meta-analysis of RCTs
    Eighteen trials with a total of 522 participants were included in this systematic review. The results showed that consumption of soy protein led to a significant decrease in total cholesterol, LDL, and proteinuria levels.
    Int Urol Nephrol, 2025 · checked 2026-09-25
17. tftm-daily-r1-12 · systematic narrative review
    However, evidence for other B vitamins (B1, B2, B3, B5, B6 and B12) is limited or inconsistent, and no human evidence exists for other vitamins. ... Despite encouraging data, there is a lack of well-controlled, large-scale human studies to validate FF as a sustainable strategy to improve vitamin status.
    Front Nutr, 2025 · checked 2026-09-25