Skip to content

Ketogenesis and beta-hydroxybutyrate

Also called: Ketosis, BHB, Ketone bodies

Written by Reviewed Sep 2026

When food runs short, the liver breaks down fat faster than it can burn the pieces. It turns the extra into small molecules called ketones and sends them out as a fuel the brain can use when blood sugar is low.

What it is

There are three ketones. Acetoacetate is the one the liver actually makes. Beta-hydroxybutyrate, usually shortened to BHB, is a close cousin of it. BHB is the main ketone in the blood during a fast, and it is the one blood meters measure. Acetone is what acetoacetate turns into when it breaks down on its own. Acetone leaves the body on the breath, which is why deep ketosis can be smelled. Breath acetone is a real marker, but it only gives a rough number.

The liver makes all of this and can use none of it. Liver cells lack an enzyme called SCOT, which every other tissue needs to burn a ketone. That one missing piece is what makes ketone making a true export system, not a pointless loop inside the liver. The liver makes a fuel it cannot use, and ships every molecule out.

BHB is not only a fuel. It also blocks a group of enzymes that help control which genes are switched on. It switches on a receptor on fat cells. And it can be stuck onto the proteins that DNA is wrapped around, as a chemical mark. Those three jobs have nothing to do with each other, and none of them needs BHB to be burned. Where they have been shown, and where they have not, is the whole point of the two sections below.

How it works

Step one is release. Insulin normally holds back fat breakdown, so when insulin falls, it lets fat breakdown happen. Three enzymes strip the fatty acids off stored fat and let them go, along with one glycerol. The fatty acids ride in the blood on a carrier protein. The glycerol travels free and is used to make new blood sugar.

Step two is the gate. The liver takes up fatty acids, but they cannot get into the mitochondria, the cell's power plants, on their own. A gatekeeper enzyme called CPT1 carries them across. A molecule called malonyl-CoA shuts that gate. Insulin raises malonyl-CoA, and the energy sensor AMPK lowers it. So when insulin falls and AMPK rises, malonyl-CoA drops, the gate opens, and fat floods into the mitochondria. This one step is where the hormones of a fast turn into real chemistry.

Step three is fat burning proper. The fatty acid chain gets chopped two carbons at a time into small pieces called acetyl-CoA. A common sixteen-carbon fat gives eight of them.

Step four is the bottleneck. Normally, acetyl-CoA pairs up with a partner molecule to enter the cell's main energy cycle. But in a fasting liver, that partner is being pulled away to make blood sugar. The liver ends up with more acetyl-CoA than it has partners for.

Step five is ketone making. The spare acetyl-CoA pieces are joined together in a few steps into acetoacetate. The key enzyme here is HMGCS2, and a fasting liver turns up its gene. Acetoacetate is then turned into BHB by an enzyme called BDH1. The balance of BHB to acetoacetate in the blood gives a direct reading of the chemical state inside the liver's mitochondria.

Step six is use. Ketones need no carrier in the blood. They cross into the brain through small doorway proteins. In any tissue that has SCOT, the steps run in reverse, and the ketone is broken back into acetyl-CoA to be burned for energy.

What switches it on

Falling insulin, rising glucagon and rising AMPK all work on the same gate, CPT1. There is also a built-in brake. BHB switches on a receptor on fat cells, called HCAR2 or GPR109A, and that slows down further fat release. In a person whose insulin works normally, ketone making limits itself.

On the timing
This is one of the few parts of fasting science where a real hour-by-hour record exists in people. It does not look like the popular version. BHB was measured every six hours in 34 adults through the standard 72 hour fast that doctors use for testing. The middle change from 12 to 18 hours was zero. Almost nothing happens to ketones in the first 18 hours. The steepest climb is between 18 and 36 hours, a middle rise of 333%. It then rose 210% from 36 to 54 hours, and 167% from 54 to 72 hours. Those are percentage changes, not blood levels. No single study in people gives actual BHB blood levels at 12, 24, 36, 48 and 72 hours. That is why the neat level-by-hour chart you see online is not repeated on this page.

What has been measured

In people

Ketones are unusually well measured in people. They are in the blood, where they can be sampled again and again. Scientists have also measured them inside the living brain, both with scanners and with thin tubes placed in the brain's blood vessels. Ketones as a fuel have been measured directly in people. Their other jobs, as signals, have not.

  • The 72 hour record: 34 adults were sampled every 6 hours through the standard fast doctors use to rule out a rare insulin-making tumour. The middle change was 0% from 12 to 18 hours, +333% from 18 to 36, +210% from 36 to 54, and +167% from 54 to 72. Doctors already use a BHB level above 2.7 mmol/L as a marker. 74% of the people whose test came back clear crossed that level before the 72 hour mark (Journal of Clinical Endocrinology and Metabolism, 2005).
  • Actual blood levels, where they were published. Total ketones, meaning acetoacetate and BHB together, were 0.20 mM after an overnight fast. They averaged 5.7 mM at rest in people fasted 3 to 5 days (American Journal of Physiology, 1983). In 13 volunteers fasting 21 days, blood ketones went from 0.1 to 6.61 mmol/L (Scientific Reports, 2024). Total ketones and BHB alone are not the same number, and the balance between them shifts as a fast gets longer.
  • How the brain's fuel changed at 3.5 days, in 9 healthy volunteers, measured two separate ways, one with a PET scanner. Sugar use in the brain's grey matter fell 26%, and in white matter it fell 27%. The fall was the same across brain areas, and blood flow to the brain did not change. The brain's uptake of BHB rose 13-fold. At 3.5 days, ketones cover roughly one quarter of the human brain's energy needs (Journal of Cerebral Blood Flow and Metabolism, 1994).
  • How ketones get into the brain, in the same 9 people. The brain's ability to take in sugar rose 55%. That was more than the lower blood sugar alone would explain, so the doorway proteins for sugar went up a little. The brain's ability to take in BHB did not rise at all. BHB flow into the brain rose more than tenfold only because blood BHB rose. So ketone entry into the human brain depends on how much is in the blood, not on extra doorways (American Journal of Physiology, 1995).
  • Inside the brain, measured with a strong MRI scanner aimed at the back of the brain. Brain BHB rose from 0.05 mmol/L before fasting to 0.60 after two days and 0.98 after three. Brain lactate rose from 0.69 to 1.47 mmol/L. Blood and brain BHB rose closely together. The brain level rose about 0.26 times as much as the blood level did (Journal of Cerebral Blood Flow and Metabolism, 2000).
  • The first finding, and it was in people. Doctors placed tubes in the brain blood vessels of three patients with obesity through 5 to 6 weeks of starvation. They showed that BHB and acetoacetate had replaced sugar as the brain's main fuel (Journal of Clinical Investigation, 1967).
  • Ketone making limits itself. The production rate starts to level off after about five days. Ketones hold back their own supply, by slowing fat release and nudging insulin up. At the same time, muscle fills up on ketones and takes less, so more of what is made goes to the brain. Under normal conditions, less than 10% of ketones leave in the urine (Diabetes/Metabolism Reviews, 1989).

In animals and in cells in a dish

Every one of BHB's jobs beyond fuel was worked out in mouse tissue, in mouse cells, or in cells grown in a dish. These are real, careful experiments. None of them has been repeated in human tissue at the level a human fast actually reaches.

  • Gene switches, in mouse tissue and cells in a dish. BHB naturally blocks one group of enzymes (class I HDACs) that keep genes switched off. In mice, giving extra BHB, fasting, and cutting calories each raised a broad gene-on mark in tissue. The genes affected were tied to resisting cell damage from free radicals. Two such genes, Foxo3a and Mt2, were switched on. Mice given BHB were strongly protected against that kind of damage (Science, 2013).
  • The HCAR2 receptor. BHB was found to be the body's own key for a receptor that was known before only as the niacin receptor. In mice, it switched on the mouse version of the receptor at levels seen in the blood during fasting. In mouse fat cells, it slowed fat release through that receptor. In a dish, it also switched on the human version of the receptor. That human test was in a dish only (Journal of Biological Chemistry, 2005).
  • A new chemical mark on DNA's packing proteins, in cells in a dish and mouse liver. BHB can be stuck onto these proteins as its own mark. Forty-four sites for it were found, about as many as for a well known gene-on mark. In a dish, high BHB raised these marks sharply. In mice, the marks rose sharply in the liver during long fasts or a drug-caused diabetic crisis. The marks sat near active genes. During starvation in mice, the rise in one mark (H3K9bhb) went along with genes that the body turns up in starvation (Molecular Cell, 2016).
  • Calming an inflammation switch, in mice and cells. BHB held back an inflammation trigger called the NLRP3 inflammasome. Acetoacetate did not, and neither did the similar short fats butyrate or acetate. BHB blocked it when it was set off by urate crystals, ATP, and harmful fats. It worked whether or not BHB was burned for fuel, and it did not rely on AMPK, free radicals, autophagy or blocked sugar burning (Nature Medicine, 2015).

Why it matters

The fuel story is settled, and it comes from people. The brain stores almost no sugar and cannot burn long fats. So it has a real problem when blood sugar runs short. Ketones solve it. They are small, they dissolve in water, they travel without a carrier, and brain cells can burn them. Three separate methods in people all show the same switch: tubes in the blood vessels, PET scans and MRI scans.

The signal story is the exciting edge of this science, and it is not yet a story about people. BHB blocking gene-switch enzymes, switching on HCAR2, and leaving its own mark on DNA's packing proteins are why ketosis gets talked about as more than an emergency fuel. Each is a real effect, shown in a real experiment. Each was shown in mouse tissue or in cells in a dish. Nobody has measured whether human tissue reaches the levels those effects need, or holds them for long enough.

A claim you will see repeated

The level-by-hour chart. Almost every fasting page gives set ketone levels, in mmol/L, by hour. The usual list is roughly 0.1 to 0.3 after an overnight fast, under 1 at 24 hours, 1 to 3 at two to three days, and 5 to 7 over weeks. Those ranges point the same way as the reviews. But the research behind this page could not trace a single one of them to a real measurement in people, so none are printed here. What is printed instead is what was actually measured: the percentage changes, the total ketone levels overnight and at 3 to 5 days, the 21 day figure, and the brain levels. Separately, ketosis and autophagy are often treated as the same event on the same clock. They are not. Ketone making is a building process in the liver's mitochondria. Autophagy is a breaking-down process in the cell's digestion sac. They share some of the same upstream signals and they rise over the same hours. But a ketone reading on a meter does not measure autophagy in any tissue.

Sources

  1. Study in people2005
    Increasing serum betahydroxybutyrate concentrations during the 72-hour fast: evidence against hyperinsulinemic hypoglycemia

    Journal of Clinical Endocrinology and Metabolism

    Doctors checked the ketone BHB every 6 hours during the standard 72 hour hospital fast used to test for insulinoma, an insulin-making tumour. 34 adults tested negative, and 21 had the tumour confirmed by surgery. In those who tested negative, ketones did not change from 12 to 18 hours, then climbed steeply. 74% of them passed 2.7 mmol/L before 72 hours.

    Beta-hydroxybutyrate measured every 6 hours through the standard 72 hour diagnostic fast in 34 adults whose fast was negative for insulinoma, alongside 21 with surgically confirmed insulinoma. Median change 0% from 12 to 18 hours, +333% from 18 to 36, +210% from 36 to 54, +167% from 54 to 72. 74% of those with a negative fast exceeded 2.7 mmol/L before 72 hours.

  2. Study in people1983
    Ketone body turnover during and after exercise in overnight-fasted and starved humans

    American Journal of Physiology

    Ketones are the fuel the liver makes from fat. After an overnight fast, blood ketones were 0.20 mM. In people who had fasted 3 to 5 days, they averaged 5.7 mM at rest.

    Total ketone bodies, acetoacetate plus beta-hydroxybutyrate, were 0.20 mM in overnight fasted subjects and averaged 5.7 mM at rest in subjects fasted 3 to 5 days.

  3. Study in people2024
    Analysis of physiological and biochemical changes and metabolic shifts during 21-Day fasting hypometabolism

    Scientific Reports

    13 volunteers followed a 34 day plan that included 21 days of eating nothing at all. Blood ketones rose from about 0.1 to about 6.61 mmol/L. The energy their bodies burned at rest fell about 20.3%. Body weight fell about 14.96%. Their bodies shifted toward burning fat.

    13 volunteers through a 34 day protocol including 21 days of complete fasting. Blood ketones rose from 0.1 plus or minus 0.04 to 6.61 plus or minus 1.25 mmol/L. Resting energy expenditure fell 20.3 plus or minus 11.13%, body weight fell 14.96 plus or minus 1.55%, and the respiratory quotient shifted toward fat.

  4. Study in people2000
    Human brain beta-hydroxybutyrate and lactate increase in fasting-induced ketosis

    Journal of Cerebral Blood Flow and Metabolism

    Healthy adults had the ketone BHB measured right inside the brain with a strong MRI scanner. Brain levels went from 0.05 mmol/L when fed to 0.60 after the second day of fasting and 0.98 after the third. Brain lactate rose from 0.69 to 1.47 mmol/L. Brain levels tracked blood levels closely.

    Beta-hydroxybutyrate measured directly in the occipital lobe of healthy adults by 4-tesla magnetic resonance spectroscopy. Brain BHB 0.05 plus or minus 0.05 mmol/L non-fasted, 0.60 plus or minus 0.26 after the second day, 0.98 plus or minus 0.16 after the third. Brain lactate rose from 0.69 plus or minus 0.17 to 1.47 plus or minus 0.22 mmol/L. Plasma and brain BHB correlated at r = 0.86, brain to plasma slope 0.26.

  5. Study in people1994
    Brain metabolism during short-term starvation in humans

    Journal of Cerebral Blood Flow and Metabolism

    9 healthy volunteers were scanned before and after 3.5 days without food. The brain's use of sugar fell 26% in grey matter and 27% in white matter, and the drop was even across the brain. Blood flow to the brain did not change. The brain's uptake of the ketone BHB rose 13-fold. By 3.5 days, ketones covered about one quarter of the brain's energy needs.

    9 healthy volunteers before and after 3.5 days of starvation. Cortical grey matter glucose metabolism fell 26% (0.294 to 0.217 micromol per gram per minute), white matter fell 27%, the fall was uniform across regions at 24 to 30%, and an independent Fick determination confirmed a 24% fall. Cerebral blood flow did not change. Global net brain uptake of beta-hydroxybutyrate rose 13-fold (0.012 to 0.155 micromol per gram per minute). At 3.5 days ketones account for approximately one quarter of cerebral energy requirements.

  6. Study in people1995
    Blood-brain barrier permeability of glucose and ketone bodies during short-term starvation in humans

    American Journal of Physiology

    The same 9 people after a short fast. Sugar got into the brain 55% more easily, more than the roughly 22% expected from lower blood sugar alone. The gate for the ketone BHB did not open wider. BHB in the blood and BHB entering the brain both rose more than tenfold. So ketones get into the brain because there is more of them in the blood, not because the brain builds more doors.

    Same 9 subjects, double indicator method. The permeability surface area product for glucose transport into brain rose 55%, more than the roughly 22% expected from the lower plasma glucose alone. The permeability surface area product for beta-hydroxybutyrate did not increase. BHB blood concentration and BHB influx into brain both rose more than tenfold, so ketone entry is governed by blood concentration rather than transporter induction.

  7. Study in people1967
    Brain metabolism during fasting

    Journal of Clinical Investigation

    Three obese patients went 5 to 6 weeks without food. Doctors sampled blood going into and out of the brain. Ketones had replaced sugar as the brain's main fuel.

    Catheterisation of cerebral vessels in three obese patients undergoing 5 to 6 weeks of starvation demonstrated that beta-hydroxybutyrate and acetoacetate had replaced glucose as the predominant fuel for brain metabolism.

  8. Review of studies1989
    Ketone body production and disposal: effects of fasting, diabetes, and exercise

    Diabetes/Metabolism Reviews

    A review. After about five days of fasting, the body's ketone output starts to level off. Ketones hold back fat breakdown and insulin, and muscle takes up less of them over time, which leaves more for the brain. Less than 10% of ketones leave in urine under normal conditions.

    Ketone body production rate begins to plateau after about five days of fasting, through ketone mediated restraint of lipolysis and insulin secretion combined with progressive saturation of muscle ketone uptake, which shunts ketones preferentially to the brain. Urinary ketone excretion remains below 10% of total turnover under physiological conditions.

  9. Animal study2013
    Suppression of oxidative stress by beta-hydroxybutyrate, an endogenous histone deacetylase inhibitor

    Science

    In mice and in a dish. The ketone BHB blocks enzymes called HDACs that help keep genes switched off. BHB, fasting and eating less each raised gene-opening marks in mouse tissue. Two protective genes, Foxo3a and Mt2, were switched on. Mice given BHB were well protected against oxidative stress, a kind of cell damage.

    Mouse tissue and cultured cells. BHB is an endogenous and specific inhibitor of class I histone deacetylases. Exogenous BHB, fasting and calorie restriction each raised global histone acetylation in mouse tissues. Foxo3a and Mt2 were induced, BHB raised histone acetylation at both promoters, both genes were activated by selective depletion of HDAC1 and HDAC2, and BHB treated mice were substantially protected against oxidative stress.

  10. Animal study2005
    (D)-beta-Hydroxybutyrate inhibits adipocyte lipolysis via the nicotinic acid receptor PUMA-G

    Journal of Biological Chemistry

    In mice and in a dish. The ketone BHB turned out to be the body's own key for the niacin receptor. At levels seen in the blood during fasting, it switched on the mouse receptor and, in a lab test, the human version. It slowed fat breakdown in mouse fat cells, and needed the receptor to do it. This loop is what keeps ketone making from running away.

    D-beta-hydroxybutyrate identified as the endogenous ligand of the niacin receptor. It activated mouse PUMA-G and, in a receptor assay, the human orthologue HM74a, at concentrations observed in serum during fasting, and inhibited lipolysis in mouse adipocytes in a PUMA-G dependent manner. This is the negative feedback loop that makes ketone production self-limiting.

  11. Animal study2016
    Metabolic Regulation of Gene Expression by Histone Lysine beta-Hydroxybutyrylation

    Molecular Cell

    In a dish and in mice. Researchers found a new tag that BHB leaves on the proteins that package DNA, at 44 spots. The tags rose sharply with high BHB in cells and in the livers of mice that fasted a long time or had a drug-triggered diabetes crisis. They sat at genes that were switched on. During starvation they rose at genes the body uses to cope with starvation.

    Cultured cells and mouse liver. Lysine beta-hydroxybutyrylation identified and verified as a distinct histone modification, with 44 sites found, a count comparable to known histone acetylation sites. Marks were dramatically induced by elevated BHB in cultured cells and in the livers of mice under prolonged fasting or streptozotocin induced diabetic ketoacidosis, were enriched at active gene promoters, and the rise in H3K9bhb during starvation was associated with genes upregulated in starvation responsive pathways.

  12. Animal study2015
    The ketone metabolite beta-hydroxybutyrate blocks NLRP3 inflammasome-mediated inflammatory disease

    Nature Medicine

    In mice and in a dish. The ketone BHB calmed an inflammation switch called NLRP3. Similar molecules did not. It worked against several triggers, including gout crystals and fatty acids. It did this by stopping potassium from leaking out of cells, and it worked without the cell burning the BHB for fuel.

    Mouse and cell systems. BHB, but not acetoacetate and not butyrate or acetate, suppressed NLRP3 inflammasome activation in response to urate crystals, ATP and lipotoxic fatty acids, by preventing potassium efflux and reducing ASC oligomerisation and speck formation. The effect was independent of chirality, and of AMPK, reactive oxygen species, autophagy and glycolytic inhibition, and occurred without BHB being oxidised.

  13. Review of studies2017
    Multi-dimensional Roles of Ketone Bodies in Fuel Metabolism, Signaling, and Therapeutics

    Cell Metabolism

    A review of ketones as both fuel and messengers. It covers the enzymes that make and use them, and why the liver cells that make ketones cannot burn them.

    Synthesis of ketone bodies as both fuel and signalling molecules, including the enzymology of HMGCS2, BDH1 and SCOT and the reason hepatocytes cannot oxidise their own product.