"Is creatine bad for you" is asked roughly 21,700 times a month in the United States, and it is almost always answered with a list of complaints — bloating, cramps, a raised blood marker — assembled from short trials. That is the wrong shape of answer. A list of symptoms tells you what was noticed; it does not tell you how many people were watched, for how long, and at what dose. This page answers the exposure question instead, because that is the question the word "bad" is really asking.
There is one trial that answers it, and it is not a sports-nutrition trial.
The largest and longest exposure, and where it came from
The National Institute of Neurological Disorders and Stroke ran Long-term Study 1 to find out whether creatine slows the progression of Parkinson disease. It randomised 1,741 men and women at 45 sites in the United States and Canada to creatine monohydrate 10 grams a day or placebo, for a minimum of five years, with a maximum follow-up of eight (Writing Group for the NINDS Exploratory Trials in Parkinson Disease, JAMA 2015;313(6):584–93, PMID 25668262).
It failed. A planned interim analysis of the 955 participants enrolled at least five years earlier found the creatine group no better than placebo on the primary global measure — summed ranks of 2,414 against 2,360, with higher meaning worse — and the trial was terminated for futility. Median follow-up at that point was four years.
That failure is why the trial matters here. It ran long, it ran large, it dosed at double the 3 to 5 grams commonly used for maintenance, and because it was a registered interventional trial it had to post its safety tables to ClinicalTrials.gov, where they sit today under NCT00449865. They cover 874 participants on creatine and 867 on placebo.
No sports-nutrition study comes close. For scale: of 359 interventional studies registered with creatine as an intervention, 30 have posted results at all.
The serious-event table
A safety reviewer reads these tables by comparing arms, not by counting events, because in a four-year trial in an older population a great many things happen to everybody.
| Creatine 10 g/d (n = 874) | Placebo (n = 867) | |
|---|---|---|
| At least one serious adverse event | 315 (36.0%) | 347 (40.0%) |
| At least one other adverse event | 698 (79.9%) | 687 (79.2%) |
Serious adverse events were less common in the group taking creatine. Non-serious events were equally common. Neither difference is the point on its own — the point is that after roughly 3,500 person-years of exposure at 10 grams a day, the arm on creatine did not fare worse than the arm on placebo.
The kidney question, answered the way it can be answered
The specific fear attached to creatine is renal. The trial recorded every serious event by term, and every renal term ran in placebo's direction or was level:
| Serious event | Creatine (n = 874) | Placebo (n = 867) |
|---|---|---|
| Renal failure | 0 | 1 |
| Renal failure, acute | 0 | 1 |
| Nephrolithiasis (kidney stones) | 2 | 6 |
| Renal colic | 0 | 2 |
| Rhabdomyolysis | 0 | 2 |
| Kidney infection | 1 | 1 |
| Pyelonephritis | 1 | 1 |
| Dehydration | 2 | 2 |
| Hepatic enzyme increased | 0 | 1 |
And in the non-serious table, proteinuria — the marker most often invoked to argue that creatine strains the kidney — appeared in 47 of 874 on creatine and 49 of 867 on placebo.
This does not prove creatine protects the kidney; with numbers this small those differences are noise in both directions. It does mean that the largest long-term dataset in existence contains no renal signal pointing at creatine at all. The separate and frequently confused question of why creatine raises the serum creatinine reading is covered in our examination of creatine and the kidneys, and the reason that blood test misleads in the first place is set out in what creatine means in bloodwork.
The two things that were more common on creatine
Reading a safety table honestly means reporting the entries that go the other way.
| Non-serious event | Creatine (n = 874) | Placebo (n = 867) |
|---|---|---|
| Muscle spasms | 101 | 53 |
| Weight increased | 137 | 101 |
| Diarrhoea | 65 | 50 |
| Nausea | 99 | 101 |
| Weight decreased | 35 | 50 |
Muscle spasms were recorded nearly twice as often on creatine. Weight gain was recorded about a third more often. Both of these are the supplement doing what it is described as doing, not a newly discovered harm — the mass is largely intracellular water and lean tissue, which is the subject of does creatine make you gain weight, and the water shift is the same one behind the complaint documented on creatine bloating. Diarrhoea was modestly more common, consistent with the gastrointestinal complaints that appear in short trials.
What 874 people over four years cannot rule out
This is the limit, and stating it is not a hedge — it is the number.
When an event occurs zero times among 874 exposed people, the conventional upper bound on its true rate is about 3 in 874, or 1 in 291. So the trial can say with confidence that creatine at 10 grams a day does not cause renal failure at a rate of, say, 1 in 50. It cannot say anything about a harm that occurs in 1 person in 5,000. No trial of this size can, and no trial of this size has been run for any supplement.
Two further limits belong here:
- The population was people with early Parkinson disease, recruited within five years of diagnosis and already on dopaminergic therapy. They are older and sicker than the average person buying creatine, which is why 36% of the placebo group had a serious event in four years. The comparison between arms is still the correct comparison — both arms had the same disease — but the absolute rates describe that group, not a gym population.
- The dose was fixed at 10 grams a day, so the tables say nothing about 20-gram loading protocols, which are shorter and are discussed on the creatine loading phase.
Three specific conditions sit outside this page's four-year safety cohort entirely — type 2 diabetes, raised blood pressure and cancer — and each has its own small and awkward literature, counted separately.
What this page does not say
It does not say creatine is proven safe, because "safe" is not a finding a trial can produce; a trial produces a bound. It does not offer any recommendation about whether to take creatine, at what amount, or for how long — that decision, and any decision involving an existing kidney, liver or cardiac condition, belongs with a clinician, per our editorial standards. The complaints people actually report in short studies are catalogued separately on creatine side effects, and the evidence for whether it does anything useful is graded endpoint by endpoint on creatine benefits.
What it does say is this: the question "is creatine bad for you" has been tested at scale, for years, at twice the usual amount, in a trial that had every incentive to find harm and reported its findings whether or not anyone read them. The answer in those tables is that the creatine arm had fewer serious events than the placebo arm, more cramps, and more weight.
Key figures
- 1,741 randomised · 874 on creatine · 867 on placebo
- 10 g/day, minimum five years planned, median four years achieved
- Serious adverse events: 36.0% creatine vs 40.0% placebo
- Renal failure: 0 vs 1 · kidney stones: 2 vs 6 · rhabdomyolysis: 0 vs 2
- Muscle spasms: 101 vs 53 · weight increase: 137 vs 101
- Rarest harm this size of trial could detect: about 1 in 291
Sources
- Writing Group for the NINDS Exploratory Trials in Parkinson Disease (NET-PD) Investigators. Effect of creatine monohydrate on clinical progression in patients with Parkinson disease: a randomized clinical trial. JAMA 2015;313(6):584–93 — PMID 25668262 · doi:10.1001/jama.2015.120
- ClinicalTrials.gov. NET-PD LS-1 Creatine in Parkinson's Disease — NCT00449865, posted results section, adverse-event tables read 2026-09-19
- ClinicalTrials.gov API v2 — census of interventional studies with creatine as an intervention, and of those with posted results, run 2026-09-19 — clinicaltrials.gov
All sources retrieved 2026-09-19. Counts are as of that date.