Pistachio basics
Are Pistachios Good for You? What the Trials Actually Fed People
We read every published human pistachio feeding trial we could find. The nutrition is solid. The dose behind the health claims is bigger than the one on the pack.
Quick answer
Yes. A 1 oz (28 g) serving of raw pistachios carries 159 calories, 5.7 g of protein, 3 g of fiber, 291 mg of potassium and mostly unsaturated fat, on USDA figures. The health trials behind the reputation are a different matter: across the 21 human trials we compiled, the 11 that fed a single fixed amount used a median of 56 g a day, about twice the labeled serving and roughly 314 calories. Eight of the 21 trials publish a funding statement we could read, and all eight name the pistachio industry.
Yes, as a food. USDA analysis puts a 1 oz serving of raw pistachios at 159 calories, 5.7 g of protein, 3 g of fiber and 291 mg of potassium, with most of the fat unsaturated. Very little argues with that.
The health claims are a separate question, and we went looking for the evidence behind them. We found 30 published human feeding trials of pistachios, reporting 21 separate studies of about 1,013 people. Reading them side by side turns up something the coverage rarely mentions: the amount of pistachio that produced the measured benefits is a lot bigger than the amount on the pack.
What is actually in a pistachio?
These are USDA's own analytical figures for raw pistachio kernels, not a brand's marketing sheet. The 1 oz column uses USDA's own portion for pistachios, which it counts as 49 kernels weighing 28.35 g.
| Per | 100 g | 1 oz (28.35 g) |
|---|---|---|
| Calories | 560 kcal | 159 kcal |
| Protein | 20.16 g | 5.7 g |
| Fat, total | 45.32 g | 12.8 g |
| of which saturated | 5.91 g | 1.7 g |
| of which monounsaturated | 23.26 g | 6.6 g |
| of which polyunsaturated | 14.38 g | 4.1 g |
| Carbohydrate | 27.17 g | 7.7 g |
| of which fiber | 10.6 g | 3.0 g |
| of which sugars | 7.66 g | 2.2 g |
| Potassium | 1025 mg | 291 mg |
| Vitamin B6 | 1.70 mg | 0.48 mg |
| Lutein + zeaxanthin | 2903 µg | 823 µg |
USDA FoodData Central, SR Legacy 170184, raw pistachios. The 1 oz column is the 100 g figure scaled to 28.35 g. Roasting and salting change the sodium and shift the water content, so a roasted pack will read slightly differently.
Two of those numbers do work the others do not. Pistachios carry more protein than most nuts. And the 2903 µg of lutein and zeaxanthin per 100 g is unusual in a nut: those are the two pigments that make up macular pigment in the retina, which is why one of the trials below is an eye study.
What have the trials actually tested?
Every published human trial we could find that fed people pistachios and measured something. We searched PubMed for randomized controlled trials and kept the ones that fed Pistacia vera kernels to humans. We dropped trials of mastic gum and Pistacia atlantica, which are different species, and mixed-nut trials, where pistachio is one component among several and nothing can be attributed to it.
| Paper | Year | n | Population | Daily dose | Duration | What it reported | Funder named |
|---|---|---|---|---|---|---|---|
| Edwards | 1999 | 10 | high cholesterol | 20% of calories | 3 wk | TC and TC/HDL fell | not readable |
| Kocyigit | 2006 | 44 | healthy | 20% of calories | 3 wk | TC fell, HDL rose | not readable |
| Sheridan | 2007 | 15 | high cholesterol | 15% of calories | 4 wk | ratios fell, TC did not | not readable |
| Gebauer | 2008 | 28 | high LDL | 32-63 and 63-126 g | 4 wk each | LDL fell 11.6% at the high dose | not readable |
| Kay | 2010 | 28 | high LDL | 32-63 and 63-126 g | 4 wk each | oxidized LDL fell | not readable |
| Li | 2010 | 59 | obesity, dieting | 53 g | 12 wk | BMI and triglycerides fell vs pretzels | not readable |
| Baer | 2012 | 16 | healthy | 42 and 84 g | 3 wk each | energy 5% below the Atwater estimate | not readable |
| Wang | 2012 | 90 | metabolic syndrome | 42 and 70 g | 12 wk | no weight gain; no group differences | Paramount Farms |
| West | 2012 | 28 | dyslipidemia | 10% and 20% of energy | 4 wk each | systolic BP fell more at the LOW dose | not readable |
| Gulati | 2014 | 60 | metabolic syndrome | 20% of energy | 24 wk | waist, glucose, LDL, hs-CRP improved | not readable |
| Kendall | 2014 | 20 | metabolic syndrome | one test meal | acute | post-meal glucose lower than bread | not readable |
| Ukhanova | 2014 | 16 | healthy | 42 and 84 g | 18 d | more butyrate-producing bacteria | not readable |
| Sauder | 2014 | 30 | type 2 diabetes | 59-128 g | 4 wk | ambulatory systolic BP fell 3.5 mm Hg | American Pistachio Growers |
| Holligan | 2014 | 28 | high LDL | 1 and 2 servings | 4 wk each | small dense LDL fell | not readable |
| Hernández-Alonso | 2014 | 54 | prediabetes | 57 g | 4 mo | glucose, insulin and HOMA-IR fell | not readable |
| Parham | 2014 | 48 | type 2 diabetes | 50 g | 12 wk | HbA1c fell 0.4 points | not readable |
| Nieman | 2014 | 19 | cyclists | 85 g | 2 wk | time-trial performance 4.8% SLOWER | American Pistachio Growers |
| Hernández-Alonso | 2015 | 54 | prediabetes | 57 g | 4 mo | small LDL particles fell | not readable |
| Kasliwal | 2015 | 60 | mild dyslipidemia | 40 g | 3 mo | HDL rose, LDL and stiffness fell | not readable |
| Sauder | 2015 | 30 | type 2 diabetes | 20% of energy | 4 wk | TC and triglycerides fell | not readable |
| Hernández-Alonso | 2017 | 49 | prediabetes | 57 g | 4 mo | two circulating microRNAs fell | not readable |
| Hernández-Alonso | 2017 | 39 | prediabetes | 57 g | 4 mo | gut-derived urine metabolites fell | not readable |
| Canudas | 2019 | 49 | prediabetes | 57 g | 4 mo | DNA oxidation fell 3.5% | Western Pistachio Association, Paramount Farms |
| Carughi | 2019 | 60 | healthy women | 56 g | 4 wk | satiety no different from biscuits | American Pistachio Growers |
| Fantino | 2020 | 60 | healthy women | 44 g | 12 wk | weight unchanged, nutrient intake rose | not readable |
| Rock | 2020 | 100 | overweight, dieting | 42 g | 4 mo | weight loss same as controls | American Pistachio Growers |
| Philpott | 2023 | 18 | athletes | 42.5 and 85 g | 2 wk each | less soreness in one leg at 85 g | not readable |
| Riley | 2024 | 66 | prediabetes | 57 g | 12 wk | no change in any glycemic marker | American Pistachio Growers |
| Scott | 2025 | 36 | low lutein intake | 57 g | 12 wk | macular pigment density rose | American Pistachio Growers |
| Ashwini | 2025 | 120 | prediabetes | 60 g | 12 wk | HbA1c fell 0.2 points | not readable |
"Not readable" means the paper sits behind a paywall and we could not read its funding statement without a subscription, so we make no claim either way about who paid for it. Full detail, including designs, registration numbers, verbatim funding statements and a source link for every row, is in the downloadable dataset.
How much is that a day?
Eleven of the 21 trials fed a single fixed amount in grams. Lined up, those doses run 40 g to 85 g a day, median 56 g, mean 54.6 g.
Fifty-six grams is almost exactly two 1 oz servings, since USDA puts a serving at 28.35 g. On USDA's calorie figure it is about 314 calories a day of pistachios, every day, for the length of the trial. The American Pistachio Growers markets a one-ounce serving of 49 kernels. The eight trials we can show it or its predecessors paid for fed between 42 g and 128 g.
The other trials say the same thing. Six set the dose as a share of daily calories, 10% to 20%, and the Penn State study reports its 10% arm as 32 g to 63 g depending on the participant. That 32 g is the smallest amount fed anywhere in the set, and it is the bottom of a range, not a fixed dose. Not one of these 21 trials tested a single 1 oz serving as its main dose. So when a headline says pistachios lower your triglycerides, the honest footnote is that nobody has checked whether the serving on the pack does it.
There is one awkward result inside that pattern. West 2012 found systolic blood pressure fell more on one serving a day than on two, which is the opposite of a dose response and which the paper reports plainly. More is not reliably better here.
Who paid for the research?
Of the 21 trials, 8 publish a funding statement we could read without a subscription. All 8 name the pistachio industry.
- Wang 2012: "This study was funded by a grant from Paramount Farms, Inc., Los Angeles, CA 90064."
- Sauder 2014: "supported primarily by the American Pistachio Growers (Fresno, CA), and, in part, by grants ... from the NIH." The paper adds that all authors received research support from APG and that APG donated the nuts.
- Nieman 2014: "This work was supported by American Pistachio Growers."
- Canudas 2019: "The Western Pistachio Association (USA) and Paramount Farms supported the trial."
- Carughi 2019: "This research was funded by the American Pistachio Growers." Three of the six authors are declared as APG consultants.
- Rock 2020: "funded by the American Pistachio Growers through a clinical trial contract to the UCSD School of Medicine."
- Riley 2024: "This trial was funded by the American Pistachio Growers."
- Scott 2025: "This project was supported by the American Pistachio Growers, who had no role in the final design, conduct, or interpretation of this study."
Two of those names need a footnote. The Western Pistachio Association renamed itself American Pistachio Growers in 2011, so the older papers and the newer ones point at the same trade body. Paramount Farms became Wonderful Pistachios and Almonds in 2015, when its parent Roll Global became The Wonderful Company.
The fair reading matters here. Industry funding is normal in nutrition research, because somebody has to buy several hundred kilos of nuts and there is no obvious public funder for a question this narrow. A funded result can still be a good one. Nieman 2014, Riley 2024 and Scott 2025 each state directly that the funder had no role in the study. Nieman's trial, paid for by APG, reported that cyclists rode slower after two weeks of pistachios. Riley's, also paid for by APG, found nothing at all. Neither is what a bought answer looks like.
The effect of the pattern is on which questions get asked. Nobody has run the trial that would test whether one 28 g serving a day changes anything, and there is no commercial reason to.
The other 13 trials sit behind paywalls. We could not read their funding statements, and we are not going to guess.
Thirty papers, twenty-one trials
Read the register straight down and it looks like thirty pieces of evidence. It is not. Thirteen of the thirty papers come from four studies.
- One Penn State dose-response study in 28 adults with high LDL produced four papers: Gebauer 2008, Kay 2010, West 2012 and Holligan 2014. Same 28 people, same three diets, same four weeks each.
- One Penn State study in 30 adults with type 2 diabetes produced two: Sauder 2014 and Sauder 2015, both registered as NCT00956735.
- One Spanish crossover trial in prediabetes produced five: Hernández-Alonso in 2014, 2015 and twice in 2017, then Canudas in 2019. The later papers analyze subsets of 49 and 39 of the original 54.
- Baer 2012 and Ukhanova 2014 report the same 16 volunteers, the same two doses and the same short feeding period, with Baer and Novotny as authors on both.
None of this is misconduct. Splitting a controlled feeding study across several outcome papers is standard practice, and each paper answers a different question. But a reader counting studies to judge how much evidence exists will overcount by about a third, and a meta-analysis that pools several papers from one trial is at risk of counting the same 28 people more than once. That risk is not theoretical: the 2022 Nutrition Reviews meta-analysis drew a published Letter to the Editor challenging exactly how it handled multiple effect sizes from single studies, and the authors reanalyzed their data in reply.
The trials are also small. Median size is 48 participants, with a range of 10 to 120.
What do the pooled results say?
Five meta-analyses have pooled this literature. Their answers are consistent and modest.
| Review | Pooled | What it found |
|---|---|---|
| Baghery 2022, Nutrition Reviews | 17 trials, 940 adults | Systolic BP -2.89 mm Hg, triglycerides -16.76 mg/dL, fasting glucose -3.62 mg/dL, HDL +1.43 mg/dL. Waist circumference, diastolic BP and BMI unchanged. |
| Hadi 2023, Crit Rev Food Sci Nutr | 12 trials | Total cholesterol -7.48 mg/dL, LDL -3.82 mg/dL, triglycerides -11.19 mg/dL. HDL unchanged. |
| Xia 2020, Medicine | 11 trials, 1593 subjects | BMI -0.18 kg/m². Body weight and waist circumference unchanged. |
| Nowrouzi-Sohrabi 2020, Diabetes Metab Syndr | 6 trials | Fasting glucose and HOMA-IR improved. HbA1c and fasting insulin did not. |
| Fogacci 2019, Crit Rev Food Sci Nutr | 4 trials | Flow-mediated dilation unchanged. Brachial artery diameter improved. |
Put in one sentence: eating roughly two servings of pistachios a day for a few months nudges blood lipids and blood pressure down by a few percent, does not change your weight, and does not consistently move HbA1c. Those are real effects and they are small ones.
What none of this shows
Nothing about heart attacks. The longest trial in the register ran 24 weeks. Every result in it is a change in a blood marker or a scan reading. No pistachio trial has ever been long enough or large enough to count a heart attack, a stroke or a death, and the register cannot tell you whether the marker changes translate into any of those.
The stronger evidence that nuts are good for you does not come from pistachio trials at all. It comes from following large populations for decades, which is what the Harvard Nutrition Source summary of nuts and heart health draws on. That work is about nuts as a category, and pistachios inherit it by membership rather than by having earned it separately.
Nothing about weight loss. Three separate lines of evidence say pistachios do not make you lose weight: Rock 2020 found identical weight loss with and without them, Fantino 2020 found no change in body weight or composition over 12 weeks, and the pooled adiposity analysis found body weight and waist circumference unmoved. What the trials do show is that adding 250 to 320 calories of pistachios a day did not make people gain weight either, because they ate less of other things. That is a genuinely useful finding, and it is not the same claim.
Nothing about roasted versus raw, or salted versus unsalted. No trial in the set compared them. If you want that difference, it is a composition question rather than a trial question, and we cover it in roasted vs raw pistachios and salted vs unsalted.
Who should be careful with pistachios
Tree-nut allergy is the one that matters. Pistachios are a tree nut and can cause severe reactions. The cashew link is the strong one: in the IDEAL study, 98 percent of cashew-sensitized children were also sensitized to pistachio, though pistachio sensitization was clinically relevant in only 34 percent of them. That gap, and the aflatoxin limits that are the only contaminant figures set for this crop, are in are pistachios poisonous.
IBS and low-FODMAP diets. Monash University rates pistachios as low FODMAP only at about a 5 g serve, seven kernels, and high FODMAP above that. Every dose in our register is 8 to 17 times that threshold. If you are following a low-FODMAP plan, the trial doses are not available to you.
Salt. Salted packs add sodium you do not control. Two trials in the register deliberately fed salted nuts, and Li 2010 compared them against salted pretzels for that reason.
Calories, on the trial doses rather than the pack. A 1 oz serving is 159 calories and slots into most days without trouble. The 56 g median trial dose is 314 calories, which is a meal component rather than a snack. If you are eating for the effects the trials measured, budget for it. More on portions in how many pistachios a day.
Does pistachio milk carry any of this over?
Barely. Pistachio milk is mostly water, so a cup carries around 2 g of protein against 5.7 g in an ounce of nuts, and almost none of the fiber. None of the trials in this register used milk. If you want the numbers side by side they are in pistachio milk vs pistachios, and the brand-by-brand panels are in pistachio milk nutrition facts. The same gap applies to pistachio butter, which at least is whole nut, and much more so to pistachio cream, where our jar audit found more sugar than nut in most jars.
Methodology, limitations and corrections
We compiled this on 7 September 2026.
How we found the trials. We searched PubMed for pistachio randomized controlled trials and read every result. We kept human feeding trials that gave participants Pistacia vera kernels and measured a health outcome. We excluded trials of Pistacia lentiscus (mastic gum), Pistacia atlantica, pistachio oil preparations, and mixed-nut interventions where pistachio was one component among several, because none of those tests the food this page is about. We read every abstract through Europe PMC and recorded the dose, duration, participant count, population and reported result as the paper states them.
How we grouped papers into trials. Where several papers report the same participants, the same doses and the same period with overlapping authors, we treat them as one trial and say so in the trial_group column of the dataset. Four such groups account for 13 of the 30 papers, and each is named in the section above.
How we read the funding. For each paper we tried to retrieve the full text through Europe PMC and read the funding, acknowledgements and conflict-of-interest sections. Eight papers, covering eight distinct trials, made a funding statement publicly readable. We quote those verbatim. For the rest we record "not readable" and draw no conclusion.
The limits are worth stating plainly:
- Absence of a readable funding statement is not absence of industry funding, and it is not evidence of independence either. Thirteen trials are unresolved, and if their statements became readable the balance could shift in either direction.
- The dose figures are not all the same kind of number. Of the 21 trials, 11 set one fixed weight in grams, 3 fed two different gram doses to the same people, 6 set the dose as a share of daily calories, and 1 fed a single test meal. Our median of 56 g uses only the 11 fixed-dose trials, which is why we say so rather than pooling all 21.
- Carughi 2019 describes its 56 g bags as in-shell. The stated energy of 1318 kJ corresponds to 56 g of kernels rather than 56 g including shell, so we have recorded it as a kernel weight and flag the discrepancy here.
- We report what each paper reports. We did not reanalyze anyone's data, assess risk of bias, or judge whether an effect is likely to be real. The "What it reported" column is a summary of the authors' own stated findings, and a positive finding in the table is not our endorsement of it.
- This is a register of trials we could find, not a systematic review. A formal review would search several databases with a registered protocol and two independent screeners. We searched one and did the screening ourselves.
- The USDA figures are for raw kernels. Roasted and salted packs will read differently, particularly on sodium.
- New trials will publish. Every figure here is stamped September 2026.
Spot an error, or want the data in another format? Email us via the about page and we will correct it and note the change here.
Citing this page
The register and the dataset are original compilations. You are welcome to reuse them, including for commercial publication, with attribution to pistachiomilk.org and a link to this page (Creative Commons CC BY 4.0). A suggested citation:
Ricci, E. (2026). Are Pistachios Good for You? A register of the human pistachio feeding trials. pistachiomilk.org. https://pistachiomilk.org/are-pistachios-good-for-you/
Version history: v1.0, 7 September 2026. First publication of the trial register (30 papers, 21 trials, literature searched 7 September 2026; funding statements read where publicly available on that date). Corrections and data updates will be listed here.
Allergy note
Pistachios are a tree nut. If you have a nut allergy, avoid them. This page describes published research and USDA composition data. It is general information, not personalized dietary or medical advice, and nothing here should be used to change treatment for diabetes, high cholesterol or blood pressure without talking to a clinician.