"Does creatine raise testosterone" is one of the most confidently answered questions in the supplement aisle, and the trials behind it are unusually easy to count. Restrict the indexed literature to randomised controlled trials that mention creatine supplementation and testosterone together and four papers come back. Read all four and the answer is the same each time, in four different ways.
The census
| Trial | n | Protocol | Testosterone result |
|---|---|---|---|
| PMID 11252073 (2001) | 11 | 20 g/day, 5 days, crossover; one heavy session | Unaffected, with or without creatine |
| PMID 14685870 (2004) | 17 | 0.3 g/kg/day, 4 wk overreaching + 2 wk taper | Fell in both groups, nadir week 3 |
| PMID 19741313 (2009) | 20 | 25 g/day × 7 d, then 5 g/day × 14 d | Unchanged at both time points |
| PMID 40265319 (2025) | 38 completers | 5 g/day, 12 weeks | Rose in both groups; no interaction |
Four trials, no increase. What follows is what each one actually did, because the differences between them are the reason the question keeps getting re-asked.
The trial everybody cites, read as written
The 2009 study is the source of nearly every claim in this area. Twenty college-aged male rugby players at a South African rugby institute completed a double-blind, placebo-controlled crossover with a six-week washout, loading 25 g of creatine a day with 25 g of glucose for seven days and then taking 5 g a day with 25 g of glucose for a further fourteen, against a glucose-only placebo. Serum testosterone and DHT were measured at baseline, day 7 and day 21.
The result in the paper's own order: "After 7 days of creatine loading, or a further 14 days of creatine maintenance dose, serum T levels did not change. However, levels of DHT increased by 56% after 7 days of creatine loading and remained 40% above baseline after 14 days maintenance (P < 0.001)." The DHT-to-testosterone ratio rose 36 per cent and then 22 per cent (p < 0.01).
So the trial found no testosterone effect and a large DHT effect. It is cited for the first as often as the second. Its authors' own proposal was that creatine "may, in part, act through an increased rate of conversion of T to DHT" — a conversion hypothesis, offered as warranting further investigation rather than as a demonstrated mechanism.
What happened to that DHT finding over the following sixteen years, what the 2025 trial found when it measured DHT and hair follicles directly, and what any of it means for hair, is the subject of this site's page tracing the hair-loss claim to its source. It is not repeated here.
The longest trial, and the number that disappears when you read both arms
The 2025 trial is the only one to run twelve weeks and the only one to measure total testosterone, free testosterone and DHT together. Forty-five resistance-trained men aged 18 to 40 were randomised to 5 g of creatine monohydrate or 5 g of maltodextrin daily, keeping their habitual diets and training; 38 completed. Blood was drawn at baseline and at twelve weeks.
There were no group-by-time interactions for any hormone (p > 0.05).
The detail underneath that sentence is the one worth carrying. Total testosterone rose in both groups over the twelve weeks:
| Group | Change in total testosterone |
|---|---|
| Creatine | +124 ± 149 ng/dL |
| Placebo | +216 ± 203 ng/dL |
Read only the creatine arm and this trial produces a headline: twelve weeks of creatine, testosterone up 124 ng/dL. Read both arms and the supplement did nothing — the group that took maltodextrin gained nearly twice as much, and the standard deviations are larger than the means in both groups. A within-group change is what happens to people over twelve weeks of training, seasons, sleep and measurement noise. The between-group comparison is the entire test, and it is the part that single-arm summaries drop.
The trial that measured seven hormones and concluded none of them mattered
The 2004 study is the most thorough endocrine assessment in the set and it is almost never quoted. Seventeen men supplemented with 0.3 g/kg/day of creatine monohydrate or placebo while resistance training five days a week for four weeks — a deliberate overreaching block — followed by a two-week taper.
Creatine did what creatine does: maximal squat and bench press fell during the early weeks on placebo but not on creatine; explosive bench-press power, body mass and leg lean body mass ended higher on creatine; the 1-RM squat improvement tended to be greater (p = 0.09).
The hormones went the other way from the folklore. Total testosterone and the free androgen index decreased in both groups, reaching a nadir at week 3, while sex hormone binding globulin moved in the opposite direction. Cortisol rose 29 per cent in the creatine group after week 1 and returned to baseline by week 2. Insulin was 24 per cent lower at week 1 and drifted back. Growth hormone and IGF-1 were unaffected.
The conclusion is stated plainly in the abstract: the performance and body-composition benefits "are not related to changes in circulating hormones." That is the strongest single statement available on this question, it comes from the trial that looked hardest for a hormonal mechanism, and it points the explanation back where the rest of the evidence puts it — at the phosphocreatine system described in what creatine is and why the size of the muscle store predicts the results.
The acute trial
The 2001 study asked whether creatine changes the hormonal response to training rather than resting levels. Eleven healthy young men completed a standardised one-hour heavy resistance session (three sets of 10RM across twelve exercises) before and after five days of 20 g/day creatine or placebo, with a five-week washout, and gave venous blood before, immediately after, and 30 and 60 minutes into recovery.
Growth hormone rose with exercise, as expected, and creatine did not alter that rise. Testosterone did not respond significantly to the session at all, with or without creatine. Cortisol during recovery tended to be higher on creatine — the same direction as the overreaching trial three years later.
The trend that became a claim
There is one result in the literature that points upward, and it is worth naming precisely because it is the one doing the work in most articles.
In a trial of ten elite rugby players designed to test whether sleep deprivation degrades a repeated passing skill, saliva was collected before each of ten trials. Salivary testosterone was unaffected by sleep deprivation itself, but trended higher with the 100 mg/kg creatine dose compared with placebo, at p = 0.067 (PMID 21324203).
Ten players. Saliva rather than serum. A single acute dose. Sleep-deprived. A p-value above the conventional threshold, reported by the authors as a trend. That study's actual finding — that a single dose of caffeine or creatine prevented the fall in passing accuracy after 3 to 5 hours of sleep — is covered in creatine and sleep.
What the evidence supports
Three statements, each of which can be checked against a named trial:
- No randomised trial has found creatine to raise testosterone. Two found it unchanged, one found it falling in both arms, and the longest found it rising more on placebo.
- One androgen did move in one trial, and it was DHT, in a pattern consistent with conversion rather than production. That result has not been reproduced under randomised conditions.
- The trial that measured the most hormones concluded that none of them explained creatine's effect, which is the same answer the mechanism gives.
None of that is advice about whether to take anything, and this page does not offer any. It is a record of what four trials measured. For the broader picture of what the same body of evidence does and does not support, the benefits graded by endpoint is the companion page, and the overall risk picture covers the safety side of the same literature.