Why your alarm stopped waking you up

Type the question into a search box and you will get the same answer everywhere: you habituated to the sound, so change your tone every few weeks.

We followed that advice back to its citations. They are a study of snooze-button presses, and a study of 21 awake men in a digger simulator. Neither one involved a sleeping person.

That does not make the advice wrong. It makes it a guess that has been repeated until it sounds like a finding — and the guess is pointing you at the wrong part of the problem.

The advice, and where its evidence actually goes

Two representative versions of it. Mudita's guide to alarm sounds says "sleep experts recommend refreshing your alarm tone every three to four weeks to keep your brain from slipping into autopilot," and — to its credit — actually cites sources for the habituation claim. Bellman's makes the same argument and cites nothing at all.

So we read Mudita's two citations. Here is what they measured.

The first is a genuinely interesting dataset and we have written about what it says about stacking alarms. It does not contain a finding about tone habituation, because it never changed anyone's tone.

The second is about habituation, properly measured — in construction equipment operators who are awake and being warned about blind spots. Whether an awake operator's response to a repeated proximity beep transfers to a sleeping person's response to a 6am alarm is exactly the question, and that paper does not answer it.

What the review literature says

This is the part that surprised us.

There is a systematic review of precisely this area: Alarm Tones, Voice Warnings, and Musical Treatments (McFarlane et al., Clocks & Sleep, 2020), covering 12 studies published between 1992 and 2020.

It does not examine habituation to repeated alarms anywhere. The studies it reviews look at acute effects — what a given sound does to you this morning — not adaptation over weeks. And on the broader question it is blunt: for adults woken abruptly, there is "insufficient evidence to support firm conclusions regarding alarm types and voice signals."

So the honest state of things is not "habituation to your alarm is a myth." It is that nobody has run the study. Thirty years of research into waking people up with sounds, and the single most repeated piece of advice about alarm tones sits in a gap in it.

What that same group did find, in a 2020 survey of 50 people, is that melodic sounds were associated with less reported grogginess than neutral ones (p = 0.023). Worth knowing — and worth saying that it is 50 people, self-reported, with no sleep staging and no reaction-time test behind it. It is a hint, not a lever.

The thing that is measured: the state, not the sound

While the tone question sits empty, the question of what state you wake into has just been answered at scale.

In January 2026, PLOS One published a nationwide study of morning sleep inertia covering 2,355 people aged 19 to 92 in South Korea. Mean duration of morning grogginess: 15.8 minutes. And the things that lengthened it:

Anxiety nearly doubling the window is the number to sit with. The authors are careful about their limits and so are we: sleep inertia was captured with a single subjective question, the design is cross-sectional so none of this proves causation, and there was no polysomnography.

Still — look at what that list is made of. Every entry is a state. They change from week to week and season to season. Your ringtone does not. If your alarm worked in June and stops working in August, the list above contains far more plausible culprits than the audio file.

The one experiment that did beat habituation

Go back to the construction study, because its actual result is the most useful thing in this whole area.

Chae et al. tested a conventional probe tone against three alternatives: a sound resembling a threat, the participant's own name, and both together. The recommendation they landed on was the own-name signal.

Read that carefully. What defeated habituation was not a different sound. It was a meaningful one. Novelty lost; salience won.

Which reframes the whole tone-rotation ritual. Cycling through six ringtones gives your brain six things that mean the same nothing. The reason your name cut through is that it carries a demand — something is being asked of you specifically — and that is a property no ringtone has, however new it is.

This is also why the fifteen seconds after the alarm matter more than the alarm. A sound can only ask. It cannot check. And a half-awake person will dismiss it with no memory of having done so, because during those fifteen seconds the part of you making decisions is not fully online yet.

What to do tomorrow

Also worth ruling out first: if your alarm is genuinely making no sound rather than being ignored, that is not habituation, it is your phone killing the alarm to save battery, and it is fixable in about four minutes.

And if you are reverse-engineering bedtimes from a cycle calculator to wake at a "good" moment, that arithmetic does not hold up either.

The short version: stop tuning the sound. It is the only variable that is easy to change, which is why it absorbs all the attention and all the blog posts, and it is the one the evidence has the least to say about.

The alarm was never the hard part. The fifteen seconds afterwards are, and no choice of ringtone has ever done anything about those.