Take 4 mg/kg of caffeine before a hard 10K and you might finish 6.8% faster. Take the exact same dose with a different gene and you finish 13.7% slower. Same drink. Same distance. The difference is one gene: CYP1A2.
Most race-week nutrition guides skip that gene entirely. They hand every athlete the same number: 3 to 6 milligrams of caffeine per kilogram of body weight, about an hour before the gun. That advice comes from real research. It’s also only half the story.
The Standard Caffeine Advice: 3–6 mg/kg, One Hour Out
The International Society of Sports Nutrition recommends 3 to 6 mg/kg of caffeine for a general performance boost. Effects show up at doses as low as 2 mg/kg. For a 70-kilogram athlete, that’s 210 to 420 milligrams, roughly two to four cups of coffee.
The timing is well studied too. Caffeine peaks in your blood 30 to 75 minutes after you drink it. That’s why “one hour before the start” became the default advice.
None of that is wrong. It’s just averaged across everybody.
Averages hide people.
What Your CYP1A2 Genotype Actually Controls
CYP1A2 is the liver enzyme that breaks down about 95% of the caffeine you drink. A single spot in that gene, called rs762551, comes in three versions: AA, AC, and CC.
Picture your liver as a drain under a sink. In the AA version, the drain runs wide open. Caffeine flows through fast, its stimulant effect timed almost perfectly to your race start, then clears out before it can work against you. In the CC version, the drain is narrow. Caffeine backs up and sits in the pipe longer than it should.
The best evidence for this comes from a 2018 trial in Medicine & Science in Sports & Exercise. Guest and colleagues gave 101 competitive cyclists 0, 2, or 4 mg/kg of caffeine before a 10-kilometer time trial, genotyped every athlete, and compared results.
AA athletes ran 4.8% faster at 2 mg/kg and 6.8% faster at 4 mg/kg (p<0.0001). AC athletes showed no real change at either dose. CC athletes were 13.7% slower at 4 mg/kg than on placebo (p=0.04).
In short: the same pre-race supplement that helps roughly half the field can hurt the other half. Researchers still don’t fully know why the CC group lost time. The leading idea is that caffeine’s stimulant load lingers longer without a matching performance payoff. The gene doesn’t just change the size of the benefit. For some athletes, it flips the sign entirely.
How Common Is Each CYP1A2 Genotype? More Than You’d Think
This isn’t a rare edge case. Using published allele frequencies for people of European ancestry, roughly 50.7% are AA, 41.0% are AC, and 8.3% are CC.
That means close to half the field gets the benefit shown in the Guest trial. Another four in ten likely feel little either way. And about one in twelve are the genotype that got measurably slower at the exact dose most pre-race advice recommends.
A 2024 study of 320 athletes found the CC genotype also gets rarer at higher competitive levels. It showed up in 18.0% of sub-elite athletes, 8.2% of elite athletes, and 0% in a small sample of Olympic-champion-caliber competitors. That’s not proof caffeine sensitivity limits careers. It’s a pattern worth watching.
The Second Gene Nobody Talks About: ADORA2A
Here’s the piece most coaching advice misses entirely. CYP1A2 controls how fast you clear caffeine. A separate gene, ADORA2A (a variant called rs5751876), controls how strongly your brain, gut, and hormones react to the caffeine still active in your system. They are not the same trait.
A 2025 study in Physiological Reports tested this directly. Athletes with the “high sensitivity” ADORA2A genotype (TT) showed a 109% jump in a gut-damage marker called iFABP after exercising with caffeine on board. Athletes with the low-sensitivity version showed only a 48% rise, which wasn’t statistically significant. CYP1A2 genotype had no measurable effect on that same gut-damage marker (p=0.701).
Translation: how fast you metabolize caffeine and how hard your gut reacts to it are controlled by two different switches. A fast metabolizer can still get hit with race-day nausea. A slow metabolizer might tolerate caffeine fine in the stomach and still run slower.
A separate 2024 trial found the same high-sensitivity ADORA2A genotype produces a bigger testosterone and growth hormone response after a race-realistic 6 mg/kg dose. Bigger hormonal response isn’t automatically a performance win. It’s another sign this gene runs its own show, independent of metabolism speed.
| CYP1A2 | ADORA2A | |
|---|---|---|
| What it governs | How fast your liver clears caffeine. | How strongly your gut, brain, and hormones react to it. |
| Risk genotype | CC (caffeine lingers, dose can backfire). | TT (gut-damage marker jumps 109% with caffeine). |
| Tested independently? | Yes, no effect on gut damage, p=0.701. | Yes, drove the entire gut-damage effect on its own. |
Where the Evidence Gets Shaky
Good science writing admits what it doesn’t know. Not every study finds a genotype effect.
A 2017 Wingate test (a short, all-out anaerobic power effort) found caffeine boosted power output in both AA and C-allele carriers, with no significant difference between groups. A 2023 meta-analysis of 12 studies backed that up: caffeine’s genotype-linked benefit shows up clearly in longer cycling time trials, where AA athletes gained roughly 5.8% (p=0.002). It essentially disappears in short power tests like a vertical jump or a Wingate sprint.
This effect looks strongest in sustained endurance efforts, not short bursts.
A 2025 systematic review pooling 19 randomized trials and 732 participants confirmed the overall pattern: AA gains, AC smaller gains, CC null or negative. But 84% of those studies were rated “some concerns” for risk of bias, and only two were rated low-risk. This is a strong signal, not settled science. Treat it as a reason to test your own response, not a guarantee of anything.
Building a Genotype-Matched Race-Day Protocol
| Genotype | Population share (European ancestry) | Guest 2018 effect at 4 mg/kg | Race-day guidance |
|---|---|---|---|
| AA (fast) | ~50.7% | +6.8% faster (p<0.0001) | Standard 3-6 mg/kg dose, well-supported by trial data. |
| AC (intermediate) | ~41.0% | No significant effect | Test in training; expect a smaller, inconsistent benefit. |
| CC (slow) | ~8.3% | -13.7% slower (p=0.04) | Skip a first-time race-day dose; stay well under 4 mg/kg and test it in training first. |
Take a cyclist I’ll call Derek, 34, chasing a sub-hour 40-kilometer time trial. He’d always followed the standard playbook: 400 mg of caffeine, about 5 mg/kg, exactly 60 minutes before every big race. Twice in a row he felt jittery, slightly nauseated, and paced worse than his training numbers predicted.
He ran his raw 23andMe data through a free rs762551 lookup tool. The result: CC, the slow-metabolizer group from the Guest trial. He wasn’t imagining the crash. He dropped his race-day dose under 2 mg/kg, tested it twice in training first, and cut 90 seconds off his next 40K with none of the nausea.
You don’t need a lab to start being smarter about this. A raw-data file from a consumer DNA test run through a free rs762551 lookup, or a sports-nutrition panel like Nutrigenomix, will tell you your CYP1A2 status. If you don’t know your genotype yet, take the safe route: the one Derek should have made from the start. Test any new caffeine dose in training, never for the first time on race morning.
This is exactly the gap AthleteOS’s race readiness check closes. Flag your CYP1A2 metabolizer status once, and your pre-race caffeine guidance stops defaulting to the population average. AA athletes see the standard dose-and-timing advice validated by the Guest trial. AC athletes get a note to test first and expect less. CC athletes get an explicit flag to avoid a first-time race-day dose. The same profile field can log ADORA2A sensitivity separately. That way a coach sees a GI or anxiety risk flag that has nothing to do with how fast an athlete clears caffeine.
Your training base still matters more than any supplement. Build it with Zone 2 work, check it with your drift ratio, and plan race week around your fitness, fatigue, and form scores. Once that’s dialed in, sign up for AthleteOS and add your CYP1A2 status to your race readiness check.
Caffeine isn’t a universal edge. For some of you, it’s a universal risk.