Modafinil, Day 5 — What The Studies Actually Say About Low Doses
I'm on modafinil again. Five days. 50mg in the morning.
Some things I've noticed so far: I feel alert. Sleep is landing better than I expected — it doesn't feel forced. Today I slept in after a long day, took 25mg instead of 50, and it was enough.
If you've never come across any of this before, here's the ground floor.
What modafinil is, plainly
Modafinil is a prescription pill originally made for narcolepsy — a condition where people fall asleep during the day involuntarily. It keeps you awake without the jittery buzz of caffeine and without the crash of amphetamines. Doctors also prescribe it for shift workers and for sleep apnoea daytime sleepiness. In the UK it's a prescription-only Schedule IV drug.
Off the prescription path, it also became known as a "smart drug" or "nootropic" — a substance people take not because they're ill but because they hope it will make them think better. The word nootropic just means "something claimed to enhance mental performance." That claim is where most of the interesting arguing happens, because it turns out to be a lot harder to prove than the marketing suggests.
Modafinil is the most-studied nootropic and the one with the strongest claim to actually doing something. The rest of this post is about what "actually doing something" means when you look at the real evidence.
The four things people report
If you read the personal accounts — forums, blogs, podcasts — the same four claims come up over and over:
- More alert. You don't feel sleepy. The urge to nap disappears.
- More focused. Tasks that felt slippery become graspable. You stay on them longer.
- More motivated. Things you were putting off suddenly seem worth doing.
- Sharper thinking. Better memory, better decisions, better problem-solving.
The first three are well-supported by the science. The fourth — actually being smarter — is the one that doesn't hold up the way people expect. Both matter, but they're different claims, and mixing them up is where the "smart drug" myth comes from.
The dose you're told to take is not the dose the evidence supports
The standard prescribed dose is 200mg once a day. That's what the drug label says. That's what a doctor writes.
The studies that actually measured whether modafinil helps rested people think better — as opposed to just keeping tired people awake — found something the marketing does not advertise: more is not reliably more. Bigger doses don't produce bigger cognitive gains. In some tests, they produce smaller ones.
- Van Puyvelde et al. 2022 (a military review of everything published on cognitive enhancement) put it plainly: "No study showed a clear dose–response effect for modafinil." Meaning: give people 100mg, 200mg, 400mg — you don't get a smooth "more drug, more benefit" line. You get a mess.
- Turner et al. 2003 (PMID 12417966, 60 healthy men): 100mg beat 200mg on a memory task (backward digit span — repeating a series of numbers in reverse). Half the dose did better than the full dose.
- Turner et al. 2014 (PMID 24306135): the more you take, the more it disrupts your sleep that night. That cost scales cleanly. The benefit does not.
- Wesensten et al. 2005: gave sleep-deprived people 100mg, 200mg, or 400mg. All three beat placebo. Higher doses were only marginally better than 100mg. Not statistically significant better — just a trend.
50mg has never been tested in a proper published cognitive study of healthy adults. Not one. The case for it is inferential — deduced, not measured. The drug company's own patent (Cephalon, US20010034373A1) claims 1-75mg is enough, but that's a commercial filing, not evidence. Volkow et al. 2009 (published in JAMA, one of the top medical journals) put people in brain scanners and found 200mg already saturates a large fraction of its target receptors — which tells you there's plenty of room to work below it.
So dropping to 25mg on a shorter day and finding it sufficient isn't defying the science. It's inside the shape of the science. The published data does not tell you 50mg — or 100mg, or 200mg — is optimal. It tells you the benefit doesn't scale cleanly with dose, and the side effects do.
Starting low is the rational default. Don't confuse that reasoning with proof.
Why your response might not match anyone else's
The most interesting recent modafinil paper isn't about average effects at all.
Van Cutsem et al. 2025 (PMID 41079714) tested the same individuals on modafinil twice, on separate days. Two things came out:
- Within the same person, the response was remarkably consistent. If it worked for you the first time, it worked the second time about the same amount. The technical measure they used was 0.90 on a scale where 1.0 is a perfect match — nearly perfect.
- Between different people, responses varied enormously. One person's "sharp and clear" is another person's "nothing happened" is another person's "wired and anxious."
Translated: modafinil is what researchers call trait-like. Your own body's response is a much better guide to what modafinil does for you than any average from a study of 60 strangers. If you take it once and get a reliable, clean, useful response, that response is likely to be there next time. If you get nothing or you get side effects, that's also likely to repeat.
This has a practical consequence. When someone says "modafinil didn't do anything for me," they might mean modafinil doesn't do anything for anyone (wrong — for many people it clearly does), or they might mean modafinil doesn't do anything for them (which for them is probably true). Both are consistent with the same data.
The catch — and it's a real one
Van Cutsem's same 2025 paper found something that everyone taking modafinil should know.
They asked participants to predict how well they were doing on a task, then measured how they actually did. On placebo (a dummy pill) the predictions matched the reality — people knew when they were sharp and when they were fading. On modafinil, the link broke. People couldn't accurately tell how well they were performing anymore.
You feel just as sharp. You might be. You might not be. Your inner sense of it has stopped being reliable.
Earlier research (Baranski & Pigeau 2004) found the same overconfidence pattern at higher doses. It's a real, documented cost of the compound, and it's the sneakiest kind, because the tool you'd normally use to check ("do I feel like I'm on top of things?") is the tool that's been unplugged.
The only guardrail is external evidence. Not "I feel sharp" — something concrete you actually did. A decision that turned out well. A piece of work you can point at. A conversation someone else confirms landed. Anything else is potentially the impairment talking about itself.
Motivation up, actual performance roughly flat
Here's where the "smart drug" myth cracks.
When you pool all the well-run studies of modafinil in rested healthy people:
- Roberts et al. 2020 (PMID 32709551) pooled 14 studies. Overall effect on thinking ability: about 0.12 on a scale where 0.2 is considered a "small" effect and 0.8 is considered a "large" effect. In other words: barely detectable. Only one specific skill — updating information in working memory — showed a real effect (0.28, small-to-moderate).
- Kredlow et al. 2019 (PMID 31433334) pooled 19 studies. Overall effect: 0.10. Same picture. Edge of detectability.
- Müller et al. 2013 compared how people enjoyed a task on modafinil versus placebo. Enjoyment shot up. Overall mood didn't change.
Put it together: people on modafinil report large gains. Studies measuring their actual performance find tiny ones. The gap is almost certainly the motivation dimension. It doesn't make you smarter — it makes you willing. Willingness gets things done that were stalled. That's genuinely useful. But it's not the same as being smarter, and calling it that is why marketing has to shout "smart drug" but the labs whisper "0.12."
What I'm watching for
- The first ten days are a honeymoon. Novelty flatters results. Real signal separates from novelty around week 3-4. Everything I've written above is inside the honeymoon window. The trait-like finding from Van Cutsem says whatever's here now should still be here at week 4 for me specifically. That's the test.
- A rare skin reaction called SJS/TEN. Very uncommon but serious. It clusters in the first 5 weeks after starting (median 17.5 days per FDA case data). Any unusual rash in that window = stop the drug immediately and see a doctor.
- Log outputs, not feelings. Because of the metacognitive-decoupling finding, the only honest record is what actually got done, not how sharp it felt.
- Weekly sleep check. Right now it's landing better than baseline, which isn't what a naïve prediction would say about a stimulant. If that flips, I'll know.
Written on the compound it describes — which per Van Cutsem is the state in which I'm least able to tell if I'm right. Cite the studies, log the outputs, don't trust the feeling.