Rosemary: the dose that speeds memory and the dose that slows it

Rosemary is a herb people use by the sprig and read about by the capsule. The one meta-analysis here pooled 45 interventional studies of 12 Mediterranean herbs and spices in people with type 2 diabetes. Five of them lowered fasting glucose: cinnamon, turmeric, ginger, black cumin and saffron, by between 27 and 17 mg/dL for the top three. Rosemary was not on that list. Absence from it can mean a weak effect or too few trials, and the review does not separate those two for this herb. What rosemary does have is a dose curve that bends back on itself. In 28 older adults with a mean age of 75, given single doses of dried rosemary powder with a week between visits, 750 mg improved speed of memory against placebo (P = 0.01) while 6000 mg made it significantly worse (P < 0.01). That was one dose per visit, measured over six hours, so it says nothing about what cooking with the herb does week after week.

The numbers, per 100 g

What cooking and storage change

• Rosemary extract added to minced meatballs slowed fat oxidation in the freezer and delayed the rancid taste, more strongly at higher doses. — minced meatballs in frozen storage, assessed by sensory panel and malonaldehyde content Evidence D source
  Delayed lipid oxidation and delayed onset of rancid taste, in proportion to the amount of extract added
• Feeding a rosemary and oregano oil mixture to quails left less lipid oxidation in their stored breast meat. — 880 one-day-old quails in 4 dietary groups over 42 days, meat sampled at several storage times (n = 880 birds) Evidence D source
  Lower thiobarbituric acid levels in raw breast meat at different storage times; the highest rosemary dose raised live weight and carcase yield

What it does to your health

• Rosemary powder acted in opposite directions by dose in older adults: 750 mg sped up memory, while 6000 mg slowed it down. — 28 older adults, mean age 75, tested 1, 2.5, 4 and 6 hours after placebo and four doses, 7-day washout (n = 28) Evidence B source
  Speed of memory significantly better on 750 mg than placebo (P = 0.01) and significantly worse on 6000 mg (P < 0.01)
• Four grams of rosemary leaf powder a day for 8 weeks added nothing to a weight loss diet in 110 patients with fatty liver disease. — 110 patients with non-alcoholic fatty liver disease, all on a weight loss diet and activity advice, 8 weeks (n = 110) Evidence B source
  No significant difference between rosemary and placebo in liver enzymes, glucose, insulin, lipids or anthropometry; only the beta-cell dysfunction index differed (p = 0.014)
• A fatty meal carrying 6 g of a 13-spice blend that included rosemary attenuated the postprandial IL-1 beta response in men with overweight or obesity, while 2 g did not. — 12 non-smoking men aged 40 to 65 with overweight or obesity and at least one cardiovascular risk factor, three-period crossover (n = 12) Evidence B source
  Significant spice-by-time interaction for IL-1 beta (P < 0.001), IL-8 (P = 0.020) and TNF-alpha (P = 0.009) in cytokine secretion from LPS-stimulated blood cells, a secondary outcome; IL-1 beta secretion fell at 6 g and not at 2 g

Safety, limits and interactions

• Across 49 studies of herbal teas including rosemary, every commonly sold tea carried measurable metals and none met the WHO limits. — 49 studies of metal concentrations in herbal teas, published 2012 to 2023 (n = 49 studies) Evidence D source
  Metals present in all commonly consumed herbal teas; arsenic and lead in tea and cadmium in black tea posed the highest risks; over 70% of the health risks judged acceptable

What people believe that the data does not support

• Of the 12 Mediterranean herbs and spices pooled for blood sugar in type 2 diabetes, 5 lowered fasting glucose and rosemary was not among them. — 77 interventional studies reviewed and 45 pooled, in people with type 2 diabetes (n = 45 studies pooled) Evidence A source
  Significant falls in fasting glucose for cinnamon, turmeric, ginger, black cumin and saffron, between 27 and 17 mg/dL for the top three; rosemary absent from the list

Who this works differently for

• In children with type 1 diabetes, a meat additive carrying rosemary extract, vitamin E and omega-3 changed the urine metabolite pattern while the clinical blood tests stayed the same. — children with type 1 diabetes and matched controls, additive delivered through meat in the standard diet (n = not stated in the abstract) Evidence B source
  Clear separation between treated and untreated diabetic children on multivariate analysis of urine metabolites, with no difference in the targeted clinical biochemistry

The bottom line

The one trial that gave rosemary every day found nothing. In 110 patients with non-alcoholic fatty liver disease, all of them on a weight loss diet and activity advice, 4 g of rosemary leaf powder a day for 8 weeks did no better than placebo on liver enzymes, glucose, insulin, lipids or body measurements. Only an index of beta-cell dysfunction differed (p = 0.014). Everything improved in both arms because both arms lost weight, which is exactly why the placebo arm was needed. The inflammation result that gets quoted is not about rosemary by itself. Twelve non-smoking men aged 40 to 65, with overweight or obesity and at least one cardiovascular risk factor, ate a fatty meal carrying a blend of 13 spices. At 6 g of blend the IL-1 beta released by their blood cells in the laboratory after the meal was lower than with no spice. The spice-by-time interactions were P < 0.001 for IL-1 beta, P = 0.020 for IL-8 and P = 0.009 for TNF-alpha. At 2 g it was not. Those were secondary outcomes measured in stimulated cells rather than anything the men felt. Rosemary was one of those 13 spices, and 6 g of blend is far more than a plate usually carries. A trial in children with type 1 diabetes used a meat additive mixing rosemary extract with vitamin E and omega-3. The urine metabolite pattern separated treated from untreated children, while the clinical blood tests showed no difference. The abstract does not say how many children took part, and a metabolite pattern is not something a child or a parent can feel. Where this herb works predictably is the storage cupboard rather than the body. Added as an extract to minced meatballs, it delayed fat oxidation and the onset of rancid taste in frozen storage, more strongly at higher doses. That is an ethanol extract used as a preservative, not the sprig and not a health effect. Fed to 880 quails over 42 days, a rosemary and oregano oil mixture left lower thiobarbituric acid levels in their stored breast meat, which is a bird and its feed rather than a kitchen. Two cautions belong here, and neither is the one people usually ask about. The first is dose. The same crossover that found 750 mg helped memory found 6000 mg made it measurably worse in adults around 75, so more is not better with this plant. Every human dose on this page is powder or extract weighed in grams, which is the supplement form of this herb rather than a sprig in a pan. The second is contamination. A review of 49 studies published between 2012 and 2023 found metals in every commonly consumed herbal tea and reported that none met the WHO requirements, while judging over 70 percent of the calculated health risks acceptable. That describes soils and supply chains across many countries rather than the plant itself. We found no study in the corpus on rosemary with medicines, with blood thinners, or in pregnancy. Those questions are open here, not answered.

Sources

1. rsmr-r3-06 · food storage experiment
   It was observed that the addition of rosemary extract delayed the oxidation of lipid fraction of products.
   Nahrung, 2000 · checked 2026-09-24
2. rsmr-r3-07 · animal experiment
   The quails fed with rations containing the VO mixture derived from rosemary and oregano had reduced thiobarbituric acid levels (TBA) in raw breast meat samples at different storage times.
   Br Poult Sci, 2012 · checked 2026-09-24
3. rsmr-r3-01 · randomised controlled trial
   There was a biphasic dose-dependent effect in measures of speed of memory: the lowest dose (750 mg) of rosemary had a statistically significant beneficial effect compared with placebo (P=.01), whereas the highest dose (6,000 mg) had a significant impairing effect (P<.01).
   J Med Food, 2012 · checked 2026-09-24
4. rsmr-r3-02 · randomised controlled trial
   However, after 8 weeks, no significant difference between the rosemary and placebo groups was detected in the variables as mentioned above except homeostasis model assessment of β-cell dysfunction (p = .014).
   Phytother Res, 2022 · checked 2026-09-24
5. rsmr-r3-04 · randomised controlled trial
   A HFCM containing 6 g spice blend attenuated HFCM-induced postprandial IL-1β secretion in men with overweight/obesity.
   J Nutr, 2020 · checked 2026-09-24
6. rsmr-r3-05 · systematic review of analytical studies
   The results indicated that all commonly consumed herbal teas included metals. None of them meet the requirements of the WHO requirements.
   Biol Trace Elem Res, 2024 · checked 2026-09-24
7. rsmr-r3-03 · systematic review and meta-analysis
   Our results showed that cinnamon, turmeric, ginger, black cumin, and saffron significantly improved the fasting glucose levels in T2DM subjects.
   Nutrients, 2024 · checked 2026-09-24
8. rsmr-r3-08 · randomised controlled trial
   After the application of the appropriate multivariate statistical tools, clear differences could be observed between treated and non-treated diabetic children and some of the metabolites associated could be identified.
   J Pharm Biomed Anal, 2010 · checked 2026-09-24