The 90-Minute Caffeine Window Most Runners Ignore
Alex Morgan
I used to think caffeine worked like a light switch. Drink it, feel it, run faster. That was the story I told myself every Saturday morning before my 10-miler around Green Lake. I would down a double espresso at 6:15 AM, lace up at 6:45, and wonder why my heart was doing the cha-cha at mile 3 while my legs felt like they were stuck in sand. The data says something different. Caffeine does not peak in your bloodstream for 45 to 60 minutes. For a CYP1A2 slow metabolizer like me, it can take 90 minutes to reach maximum plasma concentration. I was timing my pre-run caffeine like someone who had never read a pharmacokinetics chart. Which, to be fair, I had not.
Here is the thing. Most running articles tell you to drink coffee 30 minutes before a workout. That advice works for fast metabolizers. It works for people who process caffeine like their liver is running a NASCAR pit crew. For the rest of us, that 30-minute window is a lie wrapped in a sports science myth. I tested this myself. Four Saturdays in a row. Same route. Same pace target. Week 1: espresso at 6:15, run at 6:45. Average heart rate 172. Felt anxious. Week 2: espresso at 5:45, run at 7:15. Average heart rate 158. Felt smooth. Week 3: no caffeine. Average heart rate 161. Felt flat. Week 4: espresso at 5:45 again. Average heart rate 157. The pattern was not subtle. It was a billboard.
If you look at the half-life, caffeine hangs around for 5 to 7 hours in a slow metabolizer. That means my 5:45 AM espresso was still doing meaningful work at 10:45 AM. My old 6:15 AM drink was still climbing when I was already cooling down. I was paying the jitter cost without collecting the performance benefit. What most people miss is that caffeine improves endurance performance through adenosine receptor antagonism. It reduces perceived exertion. It does not make your muscles stronger; it makes your brain less willing to quit. But that effect requires the drug to actually be in your system at a sufficient concentration. If you are a slow metabolizer and you drink coffee right before a run, you are running on placebo and anxiety.
The practical fix is simple. If you are a runner and you do not know your CYP1A2 status, assume you are a slow metabolizer. About 50% of the population is. Drink your caffeine 60 to 90 minutes before you start. Not 30. Not 45. Give your liver time to do its job. I use the Caffeine Intake Calculator to track my daily total so I do not accidentally stack a pre-run dose on top of my morning habit and blow past the 400mg FDA limit. It is a simple tool, but it keeps me honest when I am tempted by a second cup.
All data stays in your browser — we never see it.
Sam, my partner, is a fast metabolizer. She can drink tea at 7 AM and feel nothing by 9. She can have a second cup at 3 PM and sleep fine. I used to be jealous. Now I am just data-driven about my envy. The Half-Life Tracker helps me visualize when I am actually caffeine-free, which matters more than I used to think. Because that 3 PM cup I used to crave? It was pushing my bedtime back by two hours without me knowing.
All data stays in your browser — we never see it.
The sleep impact is the other half of the story. When I was drinking at 6:15 AM, I thought I was safe for an 11 PM bedtime. But if the half-life is 6 hours, that means 25% of the dose is still active at 12:15 PM the next day. Not enough to feel wired, but enough to suppress deep sleep architecture. The Sleep Impact Estimator confirmed what my Oura ring was already hinting at. Actually, I rechecked that. It is 25% remaining after two half-lives, which for me is roughly 12 hours. So that 6:15 AM dose was still whispering to my brain at 10 PM. No wonder my deep sleep was thin. No wonder I woke up groggy despite 8 hours in bed.
All data stays in your browser — we never see it.
I am not saying everyone needs to wait 90 minutes. If you are a fast metabolizer, 30 minutes might be perfect. But if you are guessing, you are probably guessing wrong. The difference between a good run and a jittery slog might be nothing more than timing. Measure it. Caffeine is a tool, not a lifestyle. Measure it.
— Alex Morgan, from the loft in Seattle
Wait, let me be more precise. The 90-minute window is not universal. It depends on your CYP1A2 genotype. Fast metabolizers, the AA genotype, clear caffeine in about 3 hours. Slow metabolizers, the AC genotype, take 6 to 8 hours. The really slow ones, CC genotype, can take 10 hours. I am AC. Sam is AA. That is why she can drink espresso at 4 PM and sleep at 10 PM. If I drink espresso at 4 PM, I am still processing half of it at midnight. My sleep tracker shows it. My mood shows it. My partner shows it when I am grumpy at breakfast.
I have logged 5,000 caffeine data points over four years. Every cup, every milligram, every bedtime, every wake time. The pattern is clear. When I drink within 8 hours of bedtime, my deep sleep drops by 15%. When I wait 10 hours, it is normal. That is not anxiety. That is pharmacokinetics. And it is personal. Your genes matter. They matter more than any blog post about coffee and productivity.
Espresso, my cat, sleeps 20 hours a day. He does not drink caffeine. He is the control group. And he is winning.
