Skip to content
HabitSeek
Back to results

Eat legumes regularly

Including beans, lentils, chickpeas, peas or soy foods in meals on a regular basis, as a partial replacement for meat or refined starch.

At a glance

  • Evidence: Pooled cohorts link higher legume intake to slightly lower all-cause mortality and stroke, but not to cardiovascular, coronary or cancer mortality, and one cohort found higher cardiovascular mortality.Cardiovascular disease, stroke and all-cause mortality (observational) · Limited certainty · Observational association
  • What’s uncertain: Legume intake was generally low in these populations, which limits the range over which any dose-response can be estimated and makes the highest category a modest absolute intake.
  • First step: Use the dose the trials actually tested: a median of about 130 g a day of pulses, roughly one serving, replacing an isocaloric amount of other food rather than adding on top.
  • Before you try it: Increase the amount gradually. All precautions
Read the full takeaway

Randomized risk-factor evidence

The randomised evidence for legumes is real but narrow: adding pulses to the diet at roughly one serving a day lowers LDL cholesterol by about 0.17 mmol/L, and the same trials point to small improvements in total cholesterol, fasting glucose and insulin resistance.

Disease-rate evidence

What legumes do to disease rates is a different and much weaker question. A systematic review built for the Nordic Nutrition Recommendations pooled 31 cohorts with over two million participants and found null associations with cardiovascular disease, coronary heart disease, stroke and type 2 diabetes, concluding that the evidence was limited and no conclusion could be drawn. Where cohort analyses do find something, the effects are small (a 6% lower all-cause mortality per 50 g a day) and they are not consistent across endpoints or populations. One large federated analysis of 27 cohorts even found a weak positive association with type 2 diabetes in Europe.

Overall interpretation

The honest position is that legumes are a reasonable protein and fibre choice with a modest, well-measured effect on a surrogate marker, and that the claim they prevent heart attacks or diabetes is not established.

A bowl of cooked beansDiscussed byChristopher GardnerView statements ↓
NutritionLong-term health10–20 min · Easy · Low
Reviewed Sep 23, 2026
Save

Scientific evidence

What the research shows

Outcome assessments based on research, including its uncertainties and limitations.

Cardiovascular disease, stroke and all-cause mortality (observational)

Finding: Mixed evidenceCertainty: Limitedstudies disagree, or the effect depends on who is studied

Evidence basis: observational studies (association, not proof of cause)

Pooled cohorts link higher legume intake to slightly lower all-cause mortality and stroke, but not to cardiovascular, coronary or cancer mortality, and one cohort found higher cardiovascular mortality.

Population: Adults in prospective cohort studies: 1,141,793 participants and 93,373 deaths in the largest dose-response meta-analysis, 2,081,432 participants across 31 cohorts in the systematic review, 7,216 older Mediterranean adults at high cardiovascular risk in PREDIMED, and 807,785 adults with up to 42,473 incident diabetes cases across 27 cohorts.

Research details

Summary

A dose-response meta-analysis of 32 cohorts covering 1,141,793 participants and 93,373 deaths found that higher legume intake was associated with lower all-cause mortality (HR 0.94, 95% CI 0.91 to 0.98) and lower stroke mortality (HR 0.91, 95% CI 0.84 to 0.99), with a 6% lower risk of all-cause mortality per 50 g a day (HR 0.94, 95% CI 0.89 to 0.99). There was no significant association for cardiovascular mortality (HR 0.99, 95% CI 0.91 to 1.09), coronary heart disease mortality (HR 0.93, 95% CI 0.78 to 1.09) or cancer mortality (HR 0.85, 95% CI 0.72 to 1.01), and the certainty of evidence was judged low to moderate. A systematic review and meta-analysis of 47 studies, including 31 cohorts with 2,081,432 participants, was suggestive of null associations for cardiovascular disease, coronary heart disease, stroke and type 2 diabetes, and graded the overall evidence as limited with no conclusion. In the PREDIMED cohort of 7,216 older adults at high cardiovascular risk, the highest tertile of total legume and dried bean intake was associated with higher cardiovascular mortality (HR 1.52, 95% CI 1.02 to 2.89 and HR 2.23, 95% CI 1.32 to 3.78 respectively), while higher total legume and lentil intake was associated with lower cancer mortality (HR 0.51, 95% CI 0.31 to 0.84 and HR 0.63, 95% CI 0.40 to 0.98). A federated meta-analysis of 27 cohorts found a weak positive association between total legume intake and incident type 2 diabetes (IRR 1.02, 95% CI 1.01 to 1.04 per 20 g a day) with no association for pulses or soy. The largest cohort analysis of the combined exposure, 135,335 adults in 18 countries, found that higher total fruit, vegetable and legume intake was associated with a hazard ratio of 0.81 (95% CI 0.68 to 0.96) for total mortality, with the lowest hazard at three to four servings a day; because that exposure is combined, it cannot be attributed to legumes alone, although legume intake examined separately was inversely associated with non-cardiovascular and total mortality in the fully adjusted models.

Why this certainty

Rated Limited. The meta-analyses are large and the direction of the all-cause mortality association is consistent, but the effect is small, the exposure is uniformly low across cohorts, and the evidence was graded low to moderate by the review that produced it and limited with no conclusion by the systematic review written for national nutrition recommendations. Certainty is not higher because there is no randomised evidence for any disease endpoint and the largest Mediterranean cohort found the opposite direction for cardiovascular mortality. It is not lower because several independent cohorts and pooled analyses agree on the all-cause mortality direction.

Limitations

Legume intake was generally low in these populations, which limits the range over which any dose-response can be estimated and makes the highest category a modest absolute intake. Diet was self-reported by food frequency questionnaire, usually measured once at baseline, and the review authors note a high degree of heterogeneity for most outcomes. The PREDIMED finding of higher cardiovascular mortality in the highest legume tertile runs against the rest of the literature and is plausibly explained by residual confounding or ill health, but it has not been reconciled. The weak positive diabetes association was concentrated in European cohorts and the authors suggest it may reflect the foods eaten alongside legumes rather than the legumes themselves. Randomised trials of cardiovascular events or death do not exist for this exposure.

Effect estimate

All-cause mortality HR 0.94 (95% CI 0.91 to 0.98) for highest versus lowest legume intake, and HR 0.94 (95% CI 0.89 to 0.99) per 50 g/day increment, across 32 cohorts with 1,141,793 participants.

Sources for this outcome:

LDL cholesterol, total cholesterol, fasting glucose and insulin resistance (randomised trials)

Finding: Benefit supportedCertainty: Moderatestudies found a benefit from doing this

Evidence basis: randomised trials

Pooled randomised trials show that diets emphasising pulses at about 130 g a day lower LDL cholesterol, and separate pooling shows small reductions in total cholesterol, fasting glucose and insulin resistance.

Population: Adults in randomised controlled trials of at least three weeks duration: 26 trials with 1,037 participants in the pulse lipid meta-analysis, and 14 crossover trials with 448 participants plus one parallel trial and one non-randomised trial in the Nordic systematic review. Participants were generally healthy adults rather than people with diagnosed cardiovascular disease or diabetes.

Research details

Summary

A systematic review and meta-analysis of 26 randomised controlled trials with 1,037 participants found that diets emphasising dietary pulses at a median dose of 130 g a day, about one serving, significantly lowered LDL cholesterol compared with isocaloric control diets (mean difference -0.17 mmol/L, 95% CI -0.25 to -0.09 mmol/L). Treatment effects on apolipoprotein B and non-HDL cholesterol were not observed. The systematic review conducted for the Nordic Nutrition Recommendations pooled 14 crossover randomised trials with 448 participants and found protective effects on total cholesterol (mean difference -0.22 mmol/L), LDL cholesterol (-0.19 mmol/L), fasting glucose (-0.19 mmol/L) and HOMA-IR (-0.30), with heterogeneity high (I2 = 52% for LDL cholesterol and above 75% for the others). That review noted that the same 31 cohorts showed null associations with cardiovascular disease and type 2 diabetes, so the randomised benefit is confined to risk factors. The total cholesterol, fasting glucose and HOMA-IR estimates come from a different pooling (14 crossover randomised trials, 448 participants) than the LDL estimate (26 randomised trials, 1,037 participants); they are not a single analysis.

Why this certainty

Rated Moderate. There are more than 26 randomised trials with a consistent direction and a statistically robust pooled estimate for LDL cholesterol, which is the strongest level of evidence available for this habit and justifies a higher rating than the cohort outcomes. It is held below Strong because the trials are small, short and heterogeneous, the effect on apolipoprotein B and non-HDL cholesterol was null, the dose tested is far above typical population intake, and the outcome is a surrogate marker rather than a clinical event.

Limitations

The lipid trials were short, mostly a few weeks, and small, with a median dose of 130 g a day of pulses that is well above the intake of the cohorts that show null disease associations, so the two evidence bases are not measuring the same exposure. The apolipoprotein B and non-HDL cholesterol results were null despite the LDL result, which argues against a large or broad lipid effect. Heterogeneity was high, I2 above 75% for several outcomes, and the review that produced these estimates graded the overall evidence on disease endpoints as limited with no conclusion. These are surrogate markers: no randomised trial has tested whether eating more legumes prevents heart attacks, strokes or diabetes.

Effect estimate

LDL cholesterol mean difference -0.17 mmol/L (95% CI -0.25 to -0.09 mmol/L) with a median pulse dose of 130 g/day across 26 randomised controlled trials (n = 1,037).

Sources for this outcome:

What people say

Who discusses this?

Public statements, with positions specific to each source and topic. These views do not determine the scientific evidence assessment.

What the research shows ↑

Practical steps and precautions

How to try it

  1. Use the dose the trials actually tested: a median of about 130 g a day of pulses, roughly one serving, replacing an isocaloric amount of other food rather than adding on top. Smaller amounts have not been shown to do the same thing.
  2. Pick the swap deliberately. In the pooled trials the benefit came from replacing other foods with pulses, not from eating pulses alongside an unchanged diet.
  3. Keep the pulses as intact as possible. In the cohort analyses the strongest associations were for non-soy legumes such as beans, lentils and chickpeas rather than for soy or for legume flours, powders and protein isolates.
  4. Tinned or dried both count, and cost is low. This is one of the few dietary changes with a plausible benefit that is cheaper than the food it displaces.

Precautions

  • Increase the amount gradually. Legumes contain fermentable fibre, and the pooled trials were mostly short crossover studies in people who had agreed to eat them, so tolerability in ordinary eating is not well characterised.
  • Do not treat this as a reason to displace other established dietary components. The cohort evidence on legumes is weak, and the mortality association reported in the largest cohort analysis was for fruit, vegetables and legumes combined, which cannot be attributed to legumes alone.
  • People with kidney disease or on potassium-restricted diets should take individual advice, because legumes are high in potassium and phosphorus.
  • This habit is about legumes as food, not legume-derived supplements, proteins, flours or powders, which were excluded from the systematic review that produced the randomised estimates.

References

Legume Consumption and Risk of All-Cause and Cause-Specific Mortality: A Systematic Review and Dose-Response Meta-Analysis of Prospective Studies.

Zargarzadeh N, Mousavi SM, Santos HO, Aune D, Hasani-Ranjbar S, Larijani B, Esmaillzadeh A. · Advances in nutrition (Bethesda, Md.) · 2023

Systematic review and dose-response meta-analysis of prospective cohort studiesVerified from abstract
Open original source

Legume consumption in adults and risk of cardiovascular disease and type 2 diabetes: a systematic review and meta-analysis.

Thorisdottir B, Arnesen EK, Bärebring L, Dierkes J, Lamberg-Allardt C, Ramel A, Nwaru BI, Söderlund F, Åkesson A. · Food & nutrition research · 2023

Systematic review and meta-analysis of prospective cohort studies and randomised controlled trialsVerified from abstract
Open original source

Legume consumption and risk of all-cause, cardiovascular, and cancer mortality in the PREDIMED study.

Papandreou C, Becerra-Tomás N, Bulló M, Martínez-González MÁ, Corella D, Estruch R, Ros E, Arós F, Schroder H, Fitó M, Serra-Majem L, Lapetra J, Fiol M, Ruiz-Canela M, Sorli JV, Salas-Salvadó J. · Clinical nutrition (Edinburgh, Scotland) · 2019

Prospective cohort analysis within a randomised trial cohort (PREDIMED)Verified from abstract
Open original source

Associations of Total Legume, Pulse, and Soy Consumption with Incident Type 2 Diabetes: Federated Meta-Analysis of 27 Studies from Diverse World Regions.

Pearce M, Fanidi A, Bishop TRP, Sharp SJ, Imamura F, Dietrich S, Akbaraly T, Bes-Rastrollo M, Beulens JWJ, Byberg L, Canhada S, Molina MDCB, Chen Z, Cortes-Valencia A, Du H, Duncan BB, Härkänen T, Hashemian M, Kim J, Kim MK, Kim Y, Knekt P, Kromhout D, Lassale C, Ridaura RL, Magliano DJ, Malekzadeh R, Marques-Vidal P, Martínez-González MÁ, O'Donoghue G, O'Gorman D, Shaw JE, Soedamah-Muthu SS, Stern D, Wolk A, Woo HW, EPIC-InterAct Consortium, Wareham NJ, Forouhi NG. · The Journal of nutrition · 2021

Federated meta-analysis of prospective cohort studies using individual participant dataVerified from abstract
Open original source

Fruit, vegetable, and legume intake, and cardiovascular disease and deaths in 18 countries (PURE): a prospective cohort study.

Miller V, Mente A, Dehghan M, Rangarajan S, Zhang X, Swaminathan S, Dagenais G, Gupta R, Mohan V, Lear S, Bangdiwala SI, Schutte AE, Wentzel-Viljoen E, Avezum A, Altuntas Y, Yusoff K, Ismail N, Peer N, Chifamba J, Diaz R, Rahman O, Mohammadifard N, Lana F, Zatonska K, Wielgosz A, Yusufali A, Iqbal R, Lopez-Jaramillo P, Khatib R, Rosengren A, Kutty VR, Li W, Liu J, Liu X, Yin L, Teo K, Anand S, Yusuf S · Lancet (London, England) · 2017

Prospective cohort study (PURE) in 18 countriesVerified from abstract
Open original source

Effect of dietary pulse intake on established therapeutic lipid targets for cardiovascular risk reduction: a systematic review and meta-analysis of randomized controlled trials.

Ha V, Sievenpiper JL, de Souza RJ, Jayalath VH, Mirrahimi A, Agarwal A, Chiavaroli L, Mejia SB, Sacks FM, Di Buono M, Bernstein AM, Leiter LA, Kris-Etherton PM, Vuksan V, Bazinet RP, Josse RG, Beyene J, Kendall CW, Jenkins DJ. · CMAJ : Canadian Medical Association journal = journal de l'Association medicale canadienne · 2014

Systematic review and meta-analysis of randomised controlled trialsVerified from abstract
Open original source

Community

Discussion

Have you tried this? What helped or made it difficult?

Reader experiences are separate from the scientific evidence. Explore all discussions

No discussions yet.

Be the first to ask a question or share your experience.

Sign in to vote or join the discussion.

Related habits

Explore other habits in a shared category. Each has its own evidence and precautions.

Editorial history and review method

Revision 3 · Approved for publication · Codex (publication approval explicitly authorized by site owner; category migration)

Prepared and checked with AI. Publication approval does not imply independent expert review. Request a correction.

The change notes previously referred to miller-2017-pure-fruit-veg-legume, which was not in the file, leaving precaution 2 resting on a source no reader could see. It has now been retrieved (PURE, 135,335 adults in 18 countries) and added, and it is cited from cvd-and-mortality with its combined fruit, vegetable and legume estimate stated as combined so that it is never read as a legume-specific effect. The low-density lipoprotein outcome now states that the total cholesterol, fasting glucose and HOMA-IR estimates come from a different pooling (14 crossover trials, 448 participants) than the LDL estimate (26 trials, 1,037 participants). The Papandreou P-trend values the audit asked for are already in that source record and no change was needed. Second reviewer must re-check: (a) the audit's request to correct the accessLevel flags for thorisdottir-2023, pearce-2021 and ha-2014 — left as abstract because the full texts were not read here, and flipping fullTextAvailable would tighten the public-release gate; (b) whether the type 2 diabetes evidence should become its own outcome, since the habit's aliases and topicTags name diabetes but no outcome covers it; (c) that the PURE source id and DOI casing match the id the old change notes used. Publication release 2026-10-04: site owner authorized a bounded AI audit and release of this exact revision. See docs/release-audit-2026-10-04.md for source-check findings and limits. No independent health-professional or peer review is claimed. Category migration 2026-10-04: removed the separate goals field; categories: Nutrition, Long-term health → Nutrition, Long-term health. Research assessments, sources, guidance, precautions, and evidence review dates are unchanged. This revision awaits publication approval; no new evidence review is claimed. Publication approval 2026-10-04: site owner explicitly accepted all current category-migration revisions for deployment and requested that remaining limitations be retained in docs/content-review-backlog.md. Approval applies to this exact content revision. No new evidence review, independent expert review, or clinical validation is claimed; existing precautions and disclosures remain in effect.