For most healthy people, nutritional ketosis is not bad for the kidneys, and the measured changes — a modest dip in glomerular filtration rate, a lower urine pH, and a rise in uric acid during adaptation — are usually transient and reversible. The genuine danger lies elsewhere: pre-existing chronic kidney disease, the rare inherited transport disorder proximal renal tubular acidosis, and the fact that ketone bodies compete with creatinine for tubular secretion, which can push a lab readout up even when the kidney is functioning normally. In plain terms: ketones are not toxic to a healthy kidney, but they change what kidney tests show and they strain an already-failing one.
The mechanism: how ketone bodies reach and stress the kidney
Ketones are not a single compound but three interconverting molecules. Beta-hydroxybutyrate (BHB) dominates, at roughly 70 to 80 percent of circulating ketones during a fast. Acetoacetate runs about 15 to 20 percent, and acetone, the volatile breakdown product, accounts for the small remainder.
The kidney plays two separate roles here. It is a ketone producer during prolonged fasting, contributing perhaps 15 to 30 percent of total ketogenesis after 48 to 72 hours. It is also a ketone filter and reabsorber, handling whatever the liver releases into the blood.
BHB is a weak acid with a pKa near 4.7. At physiological blood pH of 7.4, nearly all of it exists as the BHB anion plus a hydrogen ion. The kidney must excrete that acid load to defend blood pH. It does this by generating ammonia in proximal tubule cells, then excreting ammonium. This is the same machinery that handles everyday metabolic acid production.
Here is the non-obvious part. The kidney's acid-excretion capacity is not fixed. In response to a sustained acid load it upregulates glutaminase and increases ammonia production over 3 to 7 days. That adaptation is why diet-induced ketosis produces only a mild, self-limited acid challenge in a healthy person rather than a progressive acidosis.
In plain terms: the kidney responds to ketones like it responds to any acid, by ramping up a buffering system it already owns, and the ramp-up takes days, not hours.
What the measurements show: quantified kidney effects of ketosis
The changes are real but modest in people with healthy kidneys. The table below reconciles the main numbers, all drawn from controlled feeding and ketogenic-diet studies in normal-weight and obese adults without kidney disease.
| Metric | Typical change in ketosis | Magnitude | Reversibility on return to carbs |
|---|---|---|---|
| eGFR | Small decline during adaptation | roughly 5 to 15 mL/min/1.73 m², often from the 110s toward the high 90s | Recovers within 1 to 2 weeks |
| Serum creatinine | Small rise (assay artifact) | 0.05 to 0.15 mg/dL | Falls back within days |
| Serum uric acid | Early spike, later fall | up to about 1 mg/dL rise in week 1, then decline | Normalizes by 4 to 6 weeks |
| Urine pH | More acidic | from about 6.0 toward 5.0 to 5.5 | Shifts back within days |
| Serum bicarbonate | Mild downward drift | stays above 22 mEq/L in healthy adults | Restores within days |
| Blood pressure | Often falls | 3 to 8 mmHg systolic in hypertensive adults | May persist if weight loss continues |
The eGFR dip deserves careful reading. Part of it is a genuine hemodynamic effect: lower insulin and higher glucagon reduce sodium reabsorption in the proximal tubule, which changes tubuloglomerular feedback and modestly lowers filtration. Part of it is not a filtration change at all. BHB and acetoacetate are reabsorbed in the proximal tubule and compete with creatinine for the same organic cation and anion transport pathways. When ketones are high, less creatinine is secreted into urine, so serum creatinine creeps up and the eGFR equation reports a lower number than the true filtration rate.
This matters because the eGFR equations clinicians use were calibrated in people eating mixed diets. A 2023 review of low-carbohydrate diets in kidney disease flagged exactly this artifact: creatinine-based estimates can overstate the filtration decline during ketosis, sometimes by 10 to 20 points. Cystatin C-based eGFR, which does not depend on tubular secretion, is the more honest yardstick in this setting.
Kidney stone risk: the strongest real concern
If there is one place ketones and their accompanying diet genuinely raise risk, it is stone formation. Ketogenic diets are typically low in citrate-rich foods like fruit and high in animal protein. Lower urine pH and lower urine citrate both favor calcium-based stones.
Uric acid changes cut both ways and confuse the picture. Early ketosis raises uric acid as the kidney competes between excreting ketones and uric acid at the same transporters. Over weeks, weight loss and improved insulin sensitivity usually push uric acid back down. The net stone risk in a given person therefore depends on which phase dominates and on fluid intake.
A practical figure: increasing daily urine volume toward 2.5 liters is the single most effective lever against stone recurrence. That is roughly 3 liters of fluid intake per day for most adults, more in hot climates.
When ketones are genuinely dangerous: CKD, RTA, and SGLT2 inhibitors
Chronic kidney disease changes the calculus. Once eGFR falls below about 45 mL/min/1.73 m², the kidney's capacity to generate ammonia and excrete acid is impaired. Adding a sustained ketone acid load to that reduced buffer reserve can tip a patient from compensated to overt metabolic acidosis, with bicarbonate dropping below 22 mEq/L. The 2024 KDIGO guidance on diabetes and CKD cautions against unsupervised very-low-carbohydrate diets in advanced CKD for this reason, and recommends monitoring bicarbonate if such a diet is attempted.
Proximal renal tubular acidosis (type 2 RTA) is the rare but important edge case. In this inherited or drug-induced disorder, the proximal tubule fails to reabsorb bicarbonate. Because ketone reabsorption and bicarbonate reabsorption share proximal machinery, ketosis can unmask or worsen the acidosis. Any fasting or ketogenic trial in a person with known type 2 RTA needs clinician supervision and bicarbonate monitoring.
SGLT2 inhibitors add a distinct wrinkle. These drugs lower glucose by spilling it into urine and, as a side effect, modestly raise ketone production, sometimes producing euglycemic ketoacidosis at normal glucose levels. This is a recognized safety concern per FDA labeling, and it is not a kidney-damage mechanism so much as an acid-base emergency. The takeaway: high ketones plus SGLT2 inhibitor plus illness or fasting is a flag to check ketones and seek care.
Red flags that mean stop and call a clinician
These are not "watch and wait" symptoms. In a person on a ketogenic diet, fasting protocol, or SGLT2 inhibitor, they warrant prompt medical assessment:
- Bicarbonate below 18 mEq/L on a metabolic panel, or any bicarbonate drop combined with fatigue or rapid breathing.
- Nausea, vomiting, or abdominal pain that does not resolve with fluids and food.
- Confusion, unusual drowsiness, or a fruity breath odor that worsens.
- A creatinine rise over 0.3 mg/dL in under 48 hours, or an eGFR drop that does not recover within 2 weeks.
- New flank pain, blood in urine, or passing a stone.
- In SGLT2 inhibitor users, any of the above even with normal blood glucose.
Timeline: what happens to the kidney during ketosis adaptation
The time course is the most useful part of the story, because most concerning readings are transient and resolve if the diet is sustained appropriately.
| Phase | Time frame | Kidney and lab changes |
|---|---|---|
| Ketone rise | Days 1 to 3 | BHB climbs to 1 to 3 mmol/L; urine pH falls; creatinine may start to rise from transport competition |
| Early adaptation | Days 3 to 7 | Uric acid peaks; eGFR dip is largest; mild acidosis is buffered by increased ammonia production |
| Stabilization | Weeks 1 to 3 | Ammonia production upregulates; bicarbonate stabilizes above 22 mEq/L; uric acid begins to fall |
| Full adaptation | Weeks 3 to 6 | eGFR and creatinine largely settle; uric acid normalizes in most; stone risk depends on fluid and citrate intake |
| Long-term | Months 3 to 12 | Kidney function tracks weight loss and blood pressure, not ketones per se; effects are individual |
| Reintroduction of carbs | Days 1 to 14 after | All values shift back toward baseline; creatinine falls first, eGFR recovers with it |
The reconciliation point: the 5 to 15 mL/min eGFR dip in the measurement table and the "largest dip in days 3 to 7" timeline row describe the same phenomenon, and neither contradicts the recovery rows. They are the same number seen from two angles.
Correcting the misconceptions
Three claims circulate widely and deserve correction.
First, "ketones damage the kidney directly." There is no established mechanism by which physiological ketone concentrations (1 to 3 mmol/L, up to about 5 to 7 mmol/L in prolonged fasting) are toxic to kidney cells. Diabetic ketoacidosis, where ketones can exceed 10 to 15 mmol/L alongside severe dehydration, is a different situation, and the injury there is driven largely by volume depletion and acidemia, not ketones alone.
Second, "a rising creatinine on keto means kidney damage." As the transport-competition mechanism shows, creatinine can rise without any fall in true filtration. This is a hypothesis-based explanation with strong supporting physiology, but it is not yet proven in every population; cystatin C measurement is the practical way to settle it for an individual.
Third, "keto causes kidney stones, always." The direction of risk is real, but the magnitude is modest and highly modifiable with hydration and citrate. Confusing a population-level risk with a personal certainty is the error.
What to actually do: practical steps to protect the kidney on ketones
If you are considering or already on a ketogenic or fasting protocol, these actions are concrete and cheap.
Drink toward 3 liters of fluid daily, targeting 2.5 liters of urine output, to blunt stone risk.
Ask for a cystatin C-based eGFR if creatinine rises, rather than accepting a creatinine-only estimate as final.
Eat citrate-rich low-carbohydrate vegetables and consider whether your protein sources skew heavily toward animal products; balance matters more than total protein in most people.
Get a baseline metabolic panel before starting, then repeat at roughly 2 weeks, 6 weeks, and 3 months, checking bicarbonate, creatinine, and uric acid. Get the repeat earlier if you have CKD, diabetes, or take an SGLT2 inhibitor.
Avoid combining prolonged fasting with an SGLT2 inhibitor without clinician guidance, and pause the fast or the drug during acute illness.
Limitations and what remains uncertain
The honest limits of this evidence base should be stated plainly. Most ketogenic-diet kidney studies are small, short (under 12 months), and conducted in people with normal kidney function or early metabolic disease. Long-term data in advanced CKD are sparse, and much of the guidance is extrapolated from acid-load physiology rather than direct trials. The creatinine-competition artifact is plausible and physiologically grounded, but the exact size of the effect varies by person and ketone level. Finally, individual susceptibility, including stone history, blood pressure response, and genetic variation in acid handling, means that two people on identical diets can have very different kidney readouts. A clinician who knows your labs is the right arbiter.