Millions of Americans reach for diphenhydramine, the active ingredient in Benadryl, ZzzQuil, and a dozen generic “PM” pain relievers, whenever they cannot fall asleep. It is the most widely used over-the-counter sleep aid in the country, cheap, readily available, and seemingly harmless. But a new preclinical study published as a preprint on bioRxiv this week suggests that taking diphenhydramine every night may actually be degrading sleep quality rather than restoring it, and that this effect could be especially problematic for people living with Alzheimer’s disease.
Researchers at the National Institute of Environmental Health Sciences and the University of North Carolina at Chapel Hill tested chronic diphenhydramine (DPH) dosing in a well-established mouse model of Alzheimer’s disease called the 5XFAD line. Their findings paint a troubling picture: the drug fragments deep sleep, produces a daytime sedation effect that is not restorative, and fails to correct any of the sleep deficits already present in Alzheimer’s-model mice.
What the Study Found
The team implanted continuous EEG/EMG telemetry recorders in 5XFAD mice (n=16) and wild-type controls (n=14), all female and aged five months. They recorded a 24-hour baseline, then administered 10 mg/kg of diphenhydramine each day at the start of the light cycle (ZT0) for one month, followed by a 48-hour recording during ongoing dosing.
The result that jumped out was a significant fragmentation of NREM2 sleep. To understand why that matters, it helps to know a bit about how sleep is structured. When we fall asleep, we first enter NREM1, a light, transitional stage often experienced as drifting off. The brain then moves into NREM2, a deeper and more consolidated stage characterized by sleep spindles and K-complexes, which is critical for memory processing and cognitive restoration. Deeper still is NREM3 (slow-wave sleep), and then REM, the dream stage.
In both the 5XFAD mice and the healthy controls, chronic diphenhydramine broke NREM2 into shorter bouts. Instead of long, stable stretches of deep, restorative sleep, the mice experienced fragmented NREM2 episodes that never fully consolidated. This pattern resembles what sleep researchers call sleep maintenance insomnia in humans: you may fall asleep fine, but you cannot stay in the deeper stages long enough to get the full benefits.
Additionally, the drug increased the total duration of NREM1 and REM during the active (wake) phase. In mice that are supposed to be awake and active, diphenhydramine pushed them into lighter sleep and even REM episodes. That is the animal analog of daytime drowsiness and sedation in humans. The drug was not delivering high-quality sleep at night; it was spreading low-quality sleep across the whole day.
Alzheimer’s Sleep Was Already Broken, and DPH Made It Worse
The 5XFAD mice at baseline already showed signs of sleep disruption. They spent more time in NREM1 and had shorter NREM2 and REM bouts compared to their healthy counterparts. These are the sorts of sleep architecture deficits that clinicians see in patients with Alzheimer’s disease: more fragmented sleep, less deep sleep, and more daytime napping.
One might hope that a sedating drug like diphenhydramine could at least normalize some of these deficits. It did not. Instead of rescuing the Alzheimer’s-model mice from their fragmented sleep, chronic DPH made the NREM2 fragmentation measurably worse. The drug added an extra layer of sleep disruption on top of the damage already caused by the disease.
This finding is especially concerning because diphenhydramine is a first-generation antihistamine with potent anticholinergic properties. Anticholinergic drugs block acetylcholine, a neurotransmitter essential for memory, learning, and indeed for normal sleep regulation. Long-term use of anticholinergic medications has been linked in epidemiological studies to increased dementia risk. The current study raises the mechanistic question: could chronic sleep disruption be part of the pathway connecting anticholinergic use to cognitive decline?
Limitations to Keep in Mind
This is a preprint, meaning it has not yet undergone peer review. The work is from a respected group at NIH and UNC Chapel Hill, and the methods are rigorous, but the conclusions should be read as provisional until formal review is complete.
The study used only female mice; sex differences in sleep pharmacology are well documented, and the results may not generalize to males. Mouse sleep is not human sleep. While the basic architecture of NREM and REM is conserved across mammals, the timing, duration, and functional roles differ. A dose of 10 mg/kg in mice does not translate directly to human dosing.
And importantly, this was chronic daily dosing for one month in relatively young animals. The human equivalent would be taking diphenhydramine or a DPH-containing PM product every night for years, which is exactly the pattern that many older adults follow.
The Bottom Line
Diphenhydramine works as a sedative. It will make you drowsy. But sedation is not sleep. What the brain needs is properly structured, consolidated sleep architecture that cycles through light sleep, deep NREM stages, and REM in the right proportions. This study adds to a growing body of evidence that anticholinergic sedatives may produce the appearance of sleep without delivering its substance, and that this mismatch could be particularly damaging for brains already vulnerable to neurodegenerative disease.
The next time you reach for a PM pain reliever or a nighttime cold medicine because you are having trouble sleeping, it is worth considering that the active ingredient might be part of the problem, not the solution. Better sleep hygiene, cognitive behavioral therapy for insomnia, and melatonin-based approaches may offer more benefit with less risk, especially for older adults concerned about cognitive health.
Source
Copeland MH, Youngstrom DE, Konrad KS, Diering GH, Letsinger AC, Aksu LR, Yakel JL, Cushman JD. Diphenhydramine Disrupts Sleep Architecture in 5XFAD Alzheimer’s Disease Model and Wild-Type Mice. bioRxiv 2026. DOI: 10.64898/2026.07.22.739929. Posted July 27, 2026. US Government work, CC0 license.