Full hormone-nutrition source

Foods, Supplements, and Dietary Patterns That Lower Testosterone in Adult Humans

Executive summary

The adult-human literature is much narrower than internet lists suggest. The clearest direct human signals for lowering testosterone come from glycyrrhizin-containing licorice, short-term very-high-protein low-carbohydrate diets, and, to a lesser extent, very-low-fat diets. In women with hyperandrogenic conditions such as hirsutism or polycystic ovary syndrome, spearmint tea and broader dietary modification also lower androgens, but these findings should not be generalized to eugonadal men or to women without hyperandrogenism. Chronic underfeeding or low energy availability can suppress testosterone in lean men and male endurance athletes, whereas heavy alcohol exposure tends to lower testosterone in men only at large acute doses or with chronic excess; low-to-moderate acute alcohol can behave differently. citeturn16search0turn17search15turn10view0turn11view3turn46view0turn23search1turn25view0turn38search1turn40search3turn34search0

The largest short-term declines with directly extractable human numbers are substantial enough to matter clinically. In men, a meta-analysis found that short-term high-protein low-carbohydrate diets produced an average reduction of about 5.23 nmol/L, roughly 37% of average testosterone for comparably aged men; glycyrrhizin-containing licorice lowered testosterone by about 26% after one week in a repeat male study, and an earlier pilot reported an approximately 35% decrease after seven days. By contrast, low-fat diets have smaller average effects, with pooled standardized mean differences around −0.38 for total testosterone and −0.37 for free testosterone. citeturn10view0turn10view4turn11view3turn17search15turn46view0

Clinical interpretation should be conservative. In men, morning testosterone has substantial biologic variability: healthy adults commonly show 10–15% fluctuation, and two measurements in the same person may differ by as much as 30% without implying disease. That means modest average shifts from diet may be statistically real yet clinically trivial unless they are sustained, symptomatic, and move a person toward a low range. The British Columbia protocol, for example, defines late-onset hypogonadism in men over 65 years as symptoms plus serum testosterone below 6–8 nmol/L. citeturn44view0

Just as important, several widely repeated claims do not hold up well in adult men. Updated meta-analyses and large observational studies do not show convincing lowering of testosterone from soy/isoflavones, plant-based or vegetarian diets, or flaxseed in adult men. Those items belong in a “popular but unsupported for lowering testosterone” category rather than the main list of plausible testosterone-lowering exposures. citeturn28search0turn28search15turn28search16turn42search2turn42search9turn29search0

The table above is based, row by row, on the following sources: licorice human studies and safety summaries; high-protein low-carbohydrate meta-analysis and older crossover data; low-fat diet meta-analysis and included-study characteristics; spearmint pilot and randomized trial evidence; calorie restriction and low-energy-availability reviews; alcohol review and dose-specific studies; dietary modification in women with PCOS; and the soy, plant-based, and flaxseed null human literature. citeturn16search0turn17search15turn20search0turn18view2turn18view3turn10view0turn10view4turn11view3turn13search8turn46view0turn46view3turn46view4turn47view0turn23search1turn25view0turn27view0turn38search1turn38search2turn40search3turn41search1turn41search17turn34search0turn33search2turn34search11turn43search8turn28search0turn28search15turn42search2turn29search0

How to read the evidence

Two points matter before interpreting any testosterone-lowering claim. First, total testosterone and free testosterone are not interchangeable; shifts in sex hormone-binding globulin can reduce free testosterone without moving total testosterone much. That distinction is particularly relevant for spearmint, calorie restriction, and PCOS-focused diet studies. Second, the population matters enormously: many positive data come either from healthy men, physically active men, or women with hirsutism/PCOS, not from mixed community samples. citeturn44view0turn23search1turn38search1turn43search8

The evidence base is also older and weaker than many readers expect. The low-fat meta-analysis included only 206 men across 6 studies with diet periods of 2–10 weeks, and only one study was a fully randomized AB/BA crossover trial; half of the included studies were assessed as medium risk of bias. The very-high-protein low-carbohydrate signal came from only 26 participants across 3 studies in the harmful subgroup, albeit with moderate certainty. Spearmint trials lasted only days to one month and were small. These are enough to identify plausible signals, but not enough to support sweeping generalizations. citeturn46view3turn47view0turn11view3turn25view0turn27view0

Approximate testosterone reduction where human studies reported directly comparable percentages

Short-term HP-LC ≥35% protein      ████████████████████  ~37%
Licorice 0.5 g glycyrrhizin/day    ███████████████       ~26%
High-protein vs high-carb diet     ████████████          ~21%

This chart intentionally includes only studies with directly extractable percentage-like human effects. It omits low-fat and spearmint interventions because the accessible reports summarized those effects mainly as standardized mean differences or statistical direction rather than directly comparable percentage changes. citeturn10view0turn10view4turn17search15turn13search8turn46view0

Interventions with the strongest human signal

Licorice

Among specific foods and herbs, licorice has some of the clearest direct human evidence. In healthy men, licorice providing the equivalent of about 0.5 g glycyrrhizin/day lowered serum testosterone by about 26% after one week in a repeat study; that study followed an earlier seven-man pilot that reported about a 35% decline after seven days. In the repeat male study, LH and 17-hydroxyprogesterone rose while free testosterone fell only slightly and not significantly, a pattern consistent with partial steroidogenic blockade rather than complete gonadal failure. A small study in 9 healthy women aged 22–26 years in the luteal phase also found lower serum testosterone after 3.5 g/day of a commercial licorice preparation. citeturn16search0turn17search15turn17search9turn20search0turn19search2

Mechanistically, licorice is one of the best-defined items on this list. Human and translational sources report inhibition of 17β-hydroxysteroid dehydrogenase and 17,20-lyase, enzymes involved in androgen synthesis, alongside well-known effects on 11β-hydroxysteroid dehydrogenase and cortisol/mineralocorticoid metabolism. That biologic coherence strengthens the inference that the observed testosterone reductions are real, even though the human trials were tiny and short. citeturn14search0turn17search0turn18view3

Clinically, licorice is notable because the effect size is not trivial. A roughly 26–35% decline in one week is much larger than ordinary day-to-day biologic variation in testosterone and could plausibly matter for symptomatic men or women already near a lower threshold. The main limitation is that almost all direct human data are old, short, and from healthy young adults, so long-term dose–response and recovery kinetics remain poorly characterized. citeturn17search15turn44view0

Very high protein with low carbohydrate

The best current evidence for a dietary pattern lowering testosterone comes from a 2022 systematic review and meta-analysis of 27 studies involving 309 participants, mostly physically active young men. Across all low-carbohydrate diets, resting testosterone showed a modest pooled decrease; however, the signal was concentrated in short-term high-protein low-carbohydrate diets defined as ≥35% protein, where the pooled effect was SMD −1.08, corresponding to about 5.23 nmol/L lower resting total testosterone, or roughly 37% of the mean testosterone level for comparably aged men. The certainty for this harmful short-term high-protein subgroup was rated moderate. In contrast, moderate-protein low-carbohydrate diets (<35% protein) did not show a consistent effect on resting testosterone. citeturn10view0turn10view4turn11view1turn11view3turn47view1

Older crossover data point in the same direction. In one classic experiment, testosterone concentrations in seven normal men were higher after 10 days on a high-carbohydrate diet than during a high-protein diet (468 ± 34 ng/dL vs 371 ± 23 ng/dL), a difference of roughly 21%. That small trial does not prove causality by itself, but together with the modern meta-analysis it supports a real effect of extreme protein-heavy dieting, especially in low-carbohydrate settings. citeturn13search8turn10view4

The proposed mechanism is unusual and still somewhat inferential. The meta-analysis authors argued that intakes at or above 35% of energy from protein may increase the need for urea-cycle upregulation, with lower testosterone and higher cortisol functioning as part of that adaptation. The authors also acknowledged an important confounder: very-high-protein diets often reduce spontaneous energy intake, and low energy intake is itself testosterone-lowering. Practically, this means the testosterone drop may reflect a mix of macronutrient composition and energy deficit, not protein alone. citeturn10view4turn11view1

Low-fat diets

A separate systematic review and meta-analysis evaluated the effect of low-fat versus higher-fat diets in men. Across 206 participants with a mean age of about 46 years, low-fat diets averaged about 19.5% of total energy from fat versus 39.6% on higher-fat diets, with diet periods lasting 2–10 weeks. The pooled results showed small-to-moderate reductions in total testosterone (SMD −0.38), free testosterone (SMD −0.37), urinary testosterone (SMD −0.38), and dihydrotestosterone (SMD −0.30), with no significant overall changes in LH or SHBG. citeturn46view0turn46view3turn47view0

One large cross-sectional study cited in that review found that men adhering to a low-fat diet had serum total testosterone about 32.7 ng/dL lower than men not following a fat-restricted diet. That observational result is directionally consistent with the intervention meta-analysis, but the interventional data are still the more informative evidence because they at least manipulate diet directly. citeturn46view4turn46view0

The important point here is one of magnitude. The low-fat effect is much smaller than the drops seen with licorice or the extreme high-protein low-carbohydrate subgroup. In practice, this makes low-fat eating more likely to be a contributing factor than a sole cause of clinically important hypogonadism, especially in men with borderline values or multiple other suppressive factors such as obesity, poor sleep, illness, or underfeeding. citeturn46view0turn44view0

Spearmint tea

Spearmint is the clearest tea-based antiandrogen in adult humans, but the evidence is women-specific and mostly limited to hyperandrogenic conditions. The human studies used spearmint (Mentha spicata), not peppermint (Mentha × piperita) or unspecified “mint tea.” A short 5-day pilot in women with hirsutism showed a reduction in free testosterone. A later 30-day randomized placebo-controlled trial in 42 women with PCOS and hirsutism found that drinking two cups of spearmint tea daily significantly lowered free and total testosterone and improved self-reported hirsutism, although the Ferriman–Gallwey objective hirsutism score did not significantly change over only 30 days. See Akdoğan et al. 2007 and Grant 2010. citeturn26search1turn27view0turn23search1turn25view0

That separation between biochemical effect and visible phenotype matters. Hair growth changes are slow, so a one-month reduction in circulating androgen may not yet translate into measurable objective hair-score improvement. The evidence therefore supports spearmint as a small biochemical antiandrogen adjunct, not as a rapid cosmetic treatment. citeturn25view0turn27view0

Proposed mechanisms include direct antiandrogenic effects on androgen synthesis and/or increased androgen metabolism, including a possible role for CYP3A4-mediated androgen metabolism. The specific active constituent is not established. Spearmint and peppermint have overlapping polyphenols, but their volatile profiles differ: spearmint is typically richer in carvone and often limonene, while peppermint is usually richer in menthol and menthone. That makes carvone-rich spearmint a plausible marker of the studied plant material, not proof that isolated carvone is the hormone-lowering agent. For the spearmint-versus-peppermint chemistry distinction, see the comparative GC-MS analysis by Buleandra et al.. However, the literature remains small, almost entirely in women with hirsutism or PCOS, and does not justify claiming that spearmint reliably lowers testosterone in eugonadal men. citeturn21search2turn23search1

Severe calorie restriction and chronic low energy availability

Energy deficit lowers testosterone by a different route than licorice or macronutrient manipulation. In a study of lean men practicing long-term severe calorie restriction with adequate nutrition, serum total testosterone and the free androgen index were lower, while SHBG was higher, than in exercising or Western-diet comparison groups. That pattern suggests a chronic lowering of bioavailable androgen status under sustained low energy intake. citeturn38search1turn38search2turn38search3

In male athletes, recent reviews describe a consistent link between low energy availability and reduced testosterone, lower bone density, and lower resting metabolic rate, especially in endurance sports. A 2023 mini-review reported that testosterone levels in male endurance athletes often remain within the formal “normal” range but are only about 55–85% of sedentary counterparts. That is endocrinologically meaningful even when it does not cross an absolute laboratory cutoff. citeturn39search4turn41search1turn41search18

At the same time, very short experimental underfeeding trials are mixed. A 2021 review summarized that 4 days of low energy availability at 15 kcal/kg lean body mass/day did not reduce resting testosterone in one study of exercising men, whereas newer work suggests that about 3 days around 19 kcal/kg LBM/day may do so, and a later review proposed that many adverse effects cluster in the range of roughly 9–25 kcal/kg fat-free mass/day. The take-home message is that chronic underfueling is the more reliable human testosterone-lowering exposure than a few days of dieting. citeturn40search1turn40search3turn41search17

Heavy alcohol exposure

Alcohol belongs in this report only with a strong dose-response caveat. A 2023 review concluded that low-to-moderate acute alcohol can increase testosterone transiently in men, whereas large acute volumes and especially chronic excess lower testosterone production. An older experimental study similarly concluded that heavy acute drinking decreases blood testosterone in men via a testicular-level effect. A 2022 Korean study also linked heavy drinking greater than 8 drinks per week, in men who flush, to a higher risk of testosterone deficiency. citeturn34search0turn34search1turn33search2turn34search11

This means alcohol is not a reliable or interpretable “testosterone-lowering food.” Rather, it is a dose-dependent endocrine toxin whose adverse effect emerges mainly with large or chronic exposure. Cross-sex generalization is especially unsafe: observational data in women have shown positive associations between alcohol intake and testosterone, so the testosterone-lowering conclusion applies mainly to heavy exposure in men. citeturn34search0turn31search22

Dietary improvement in women with PCOS

Women with PCOS are a distinct clinical category because reducing elevated androgens is typically therapeutic, not harmful. A 2024 review summarized a meta-analysis of 20 randomized controlled trials involving 1113 women with PCOS, showing that dietary modification improved clinical hyperandrogenism, increased SHBG, and reduced total testosterone and free androgen index. citeturn43search8

This finding matters because it shows that some “testosterone-lowering” dietary effects can be beneficial in the right endocrine context. It should not be generalized to adult men, nor to women without hyperandrogenism. In men with obesity-related low testosterone, the opposite often happens: treating the underlying metabolic condition and producing substantial weight loss can actually raise testosterone. citeturn43search8turn44view0

Mechanisms

Across these interventions, the recurring biologic themes are direct steroidogenic enzyme inhibition, altered androgen binding or metabolism, hypothalamic-pituitary-gonadal suppression during energy stress, and testicular toxicity or stress signaling at high alcohol exposure. The diagram below is a simplification of the best-supported pathways from the human literature and closely related mechanistic work. citeturn18view3turn21search2turn10view4turn39search4turn34search0turn43search8

flowchart LR
    A[Licorice glycyrrhizin] --> B[Inhibits 17β-HSD and 17,20-lyase]
    B --> C[Lower androgen synthesis]

    D[Spearmint tea] --> E[Antiandrogenic effect on androgen synthesis or metabolism]
    E --> F[Lower free and sometimes total testosterone]

    G[Very high protein plus low carbohydrate] --> H[Higher cortisol and urea-cycle demand]
    H --> I[Adaptive reduction in resting testosterone]

    J[Severe calorie restriction or chronic low energy availability] --> K[Lower leptin insulin and T3]
    K --> L[HPG-axis downregulation and higher SHBG]
    L --> M[Lower bioavailable androgen]

    N[Heavy alcohol exposure] --> O[Leydig-cell dysfunction plus oxidative and HPA stress]
    O --> P[Lower testosterone with heavy or chronic use]

    Q[Dietary improvement in PCOS] --> R[Improved insulin sensitivity and higher SHBG]
    R --> S[Lower ovarian androgen output]

For licorice, the mechanism is relatively direct: the human literature consistently points to inhibition of 17β-HSD and 17,20-lyase. For spearmint, the mechanism is less settled, but the main hypotheses are direct interference with androgen synthesis and increased androgen metabolism. For chronic underfeeding, the mechanism is more systemic: lower leptin, insulin, and triiodothyronine, plus altered gonadotropin signaling, shift the body away from reproduction and toward energy conservation. The very-high-protein low-carbohydrate pattern may act partly through higher cortisol and a need to support nitrogen disposal, whereas heavy alcohol exposure appears to act through Leydig-cell dysfunction, oxidative stress, inflammation, and HPA-axis activation. In PCOS-focused diet trials, the likely androgen-lowering route is better insulin sensitivity and higher SHBG, which reduce free androgen exposure and ovarian androgen drive. citeturn18view3turn21search2turn10view4turn39search4turn40search3turn34search0turn43search8

Safety, interactions, and clinical relevance

The main clinical question is not simply whether testosterone moved, but whether it moved enough to matter. In men, morning testosterone commonly varies by 10–15%, and two measurements may differ by up to 30% in the same individual. Against that background, the average shifts seen with low-fat diets are often likely to be small in clinical terms, especially if a person starts in the mid-normal range. The larger drops reported with licorice and very-high-protein low-carbohydrate diets, by contrast, exceed ordinary biologic fluctuation and are much more likely to be clinically relevant if sustained. citeturn44view0turn17search15turn10view4turn46view0

Licorice carries the most important safety burden. The active concern is glycyrrhizin-containing licorice, not every candy marketed as “licorice,” because many commercial “licorice” products in the United States do not contain actual licorice root. Glycyrrhizin exposure can produce hypertension, hypokalemia, sodium retention, arrhythmia, pseudohyperaldosteronism, and clinically relevant interactions with cardiac glycosides, diuretics, warfarin, cyclosporine, and corticosteroids. Heavy use in pregnancy is unsafe. These safety issues are severe enough that licorice should be treated as a potential cause of otherwise unexplained low testosterone, low potassium, or hypertension, not as a benign dietary experiment. citeturn18view2turn18view3

Spearmint tea is lower risk than licorice, but caution is still warranted. Mint can worsen gastroesophageal reflux, and high-dose concentrated spearmint products are not well studied in people with liver or kidney disease or in pregnancy and breastfeeding. The positive androgen-lowering studies used tea, not concentrated extracts, and they were conducted in women with hyperandrogenic conditions. For a practical reading of the evidence, “spearmint” should mean 100% spearmint leaf / Mentha spicata; products labeled only as mint, peppermint, or mint blend should not be assumed to match the studied intervention. citeturn24news28turn24news29

For dietary patterns, context is everything. The evidence does not support saying that all low-carbohydrate diets lower testosterone; the signal is concentrated in extreme high-protein versions. In fact, the meta-analysis explicitly notes that there were no clear adverse endocrine effects in low-carbohydrate diets using around 30–31% protein, and that most free-living low-carbohydrate diets likely remain below the problematic threshold. Likewise, low-fat diets show only modest average effects and should be thought of as one modifiable contributor among many. citeturn10view4turn11view1

For calorie restriction and low energy availability, the safety story is broader than testosterone alone. Chronic underfueling is linked to bone impairment, lower resting metabolic rate, reduced libido, fatigue, and the broader syndrome of relative energy deficiency in sport. In athletes and lean adults, the goal should be restoring adequate energy availability, not treating the testosterone number in isolation. Conversely, in men with obesity and type 2 diabetes, significant weight loss can increase testosterone and improve symptoms, so clinicians must distinguish therapeutic weight loss in obesity from harmful chronic underfeeding in lean individuals. citeturn39search4turn41search18turn44view0

For women, the practical threshold question is different. The British Columbia protocol notes that the upper limit of normal total testosterone in women is roughly 2–3 nmol/L, and levels greater than about 5 nmol/L or rapid virilization suggest an alternate etiology such as an androgen-secreting tumor. That means self-treatment with licorice, teas, or “PCOS diets” should never substitute for proper endocrine evaluation when androgen excess is marked or rapidly progressive. citeturn44view0

Practical guidance and research gaps

Practical guidance

For clinicians or individuals evaluating unexpectedly low testosterone, the most useful history questions are surprisingly concrete: Has the person been consuming real licorice or glycyrrhizin-containing supplements? Are they on a cutting diet that is both low-carbohydrate and extremely high in protein? Has dietary fat fallen into the very-low range? Are they chronically underfueled because of dieting or endurance training? Is alcohol intake heavy or escalating? Those are higher-yield questions than broad concern about soy, flax, or a vegetarian diet. Any abnormal testosterone result should be repeated under proper conditions—morning, ideally fasting, and using the same laboratory—before concluding that a food or supplement truly caused a clinically meaningful decline. citeturn18view2turn10view4turn46view0turn39search4turn34search0turn44view0

Mitigation generally follows directly from mechanism. If licorice is the likely culprit, the practical step is to remove glycyrrhizin-containing products and review interacting medications. If an athlete or bodybuilder is using an extreme protein-heavy low-carbohydrate regimen, the evidence suggests lowering protein below the ≥35% of energy range implicated in the suppressive studies and correcting any associated energy deficit. If chronic underfueling is present, the priority is restoring adequate calories and screening for RED-S rather than chasing testosterone alone. For borderline-low testosterone in men, it is also reasonable to reconsider long-running very-low-fat diets. citeturn18view3turn10view4turn11view1turn39search4turn46view0

For women with hirsutism or PCOS, the human data support spearmint tea and general dietary improvement only as adjuncts with modest biochemical effects. The tea evidence is specifically for Mentha spicata, so peppermint and generic mint blends should be treated as unproven for this purpose. If androgen excess is severe, rapidly progressive, or accompanied by virilization, more serious causes must be excluded first. citeturn25view0turn43search8turn44view0

Items often claimed to lower testosterone but not convincingly supported

The most important “avoid over-interpreting internet lists” point concerns soy. The 2021 updated meta-analysis of clinical studies found that, regardless of dose and duration, neither soy protein nor isoflavone exposure altered total testosterone, free testosterone, or estradiol in men. Large observational analysis of plant-based dietary content has also found no association between plant-based diet indices and serum testosterone in U.S. adult men. citeturn28search0turn28search15turn42search2turn42search9

Flaxseed is similar. A 2023 meta-analysis concluded that flaxseed supplementation had no significant effect on sex hormones in adults, despite the persistence of mechanistic speculation and scattered case reports. Likewise, a 2023 review of vegetarian diets found no conclusive overall negative associations with male sex hormone levels. These items should therefore be treated as not convincingly testosterone-lowering in adult men at present. citeturn29search0turn42search7

Research gaps

The field has four major weaknesses. First, many of the most memorable positive human studies—especially for licorice and spearmint—are very small, old, and short-term. Second, several dietary analyses rely on crossover designs with limited randomization or on athletic populations that are not representative of the broader adult population. Third, human data are strongly sex- and condition-specific, with positive signals in men for some exposures and in women with PCOS/hirsutism for others, leaving large gaps for women without hyperandrogenism and for older, comorbid adults. Fourth, many studies report only biochemical changes, not whether the change was accompanied by clinically meaningful shifts in symptoms, fertility, body composition, or bone health. citeturn46view3turn11view3turn27view0turn39search4

The most useful next-generation studies would be larger randomized trials, modern LC-MS/MS-based hormone assays, explicit measurement of free testosterone or reliable calculated bioavailable testosterone, careful tracking of energy availability, and direct comparison of macronutrient composition versus energy deficit. Until such data exist, the evidence supports a short, practical conclusion: in adult humans, the most credible diet-related testosterone-lowering exposures are glycyrrhizin-containing licorice, extreme short-term very-high-protein low-carbohydrate dieting, chronic underfueling, and heavy alcohol excess, while claims about soy, plant-based eating, and flaxseed lowering testosterone in men are not well supported by current human evidence. citeturn18view3turn10view4turn46view0turn39search4turn34search0turn28search0turn29search0