Two glasses of wine with dinner on Friday. Saturday morning, your Oura sleep score shows 88. Great sleep, longer deep sleep, minimal wake time.
No alcohol on Tuesday night. Wednesday morning, the same device shows 74. Restless night, less deep sleep, 22 minutes of wake time.
Wait. Drinking improved your sleep? That’s not what every sleep scientist on every podcast says.
Welcome to one of the stranger artifacts in consumer sleep tracking. The sleep score can absolutely go up when you drink, for specific and misleading reasons, and understanding why tells you a lot about what these scores actually measure.
What a Sleep Score Is Made Of
Most consumer sleep scores combine several inputs into a single number. The weighting differs by device, but the ingredients are similar.
Total sleep duration. How long you were asleep.
Sleep stage distribution. How much time you spent in deep sleep, REM, and light sleep. Typically scored against adult norms.
Sleep efficiency. The ratio of time asleep to time in bed. Higher is better.
Wake events. How often you woke during the night. Fewer and shorter is better.
Resting heart rate during sleep. Lower (relative to your baseline) is scored as better recovery.
HRV during sleep. Higher (relative to your baseline) is scored as better recovery.
The score blends these into one number between 0 and 100 that’s supposed to represent how well you slept.
The problem is what alcohol does to these measurements.
What Alcohol Actually Does to Sleep
Alcohol has a well documented and surprisingly consistent effect on sleep architecture.
Sleep onset is faster. Alcohol is a sedative. You fall asleep quicker, often within 10 minutes of lying down versus 15 to 25 minutes sober. This looks like “better sleep efficiency” on a wearable.
Deep sleep is compressed into the first sleep cycle. Instead of deep sleep being distributed across the first half of the night, alcohol front loads it. You get a lot of deep sleep in the first 2 to 3 hours. The device sees a higher deep sleep duration total.
REM sleep is suppressed. Alcohol strongly reduces REM sleep, particularly in the first half of the night. This is a well documented consistent finding. REM rebound happens in the second half if the alcohol clears in time.
Heart rate stays elevated. Alcohol raises heart rate for several hours while your liver metabolises it. A glass of wine at 8pm means your sleeping heart rate at midnight is 5 to 10 beats above baseline. This is measurable and should score as worse recovery.
HRV drops significantly. Alcohol is one of the most reliable HRV suppressants. A moderate drinking session will drop overnight HRV by 10 to 20 percent. This is one of the clearest physiological signals you have.
Wake events in the second half of the night increase. As alcohol metabolises, the depressant effect clears. You become more easily arousable. Often wake up at 3am or 4am.
Total sleep duration often decreases slightly. Most people sleep 10 to 30 minutes less after moderate drinking than they would sober.
How This Shows Up in the Score
Here’s where the scoring gets weird.
The sleep score weighted heavily on sleep efficiency and deep sleep duration will often go up after moderate drinking. Faster sleep onset plus front loaded deep sleep plus the compression of the total sleep window can actually improve both of those metrics in the device’s view.
Meanwhile, the metrics that should penalise alcohol are either downweighted or invisible to the score.
Elevated heart rate during sleep. This is captured by most devices but contributes less to the composite score than people assume. Whoop’s recovery score, calculated after sleep, will reflect it. The sleep score itself often doesn’t penalise it heavily.
REM suppression. Devices measure REM, but REM duration is weighted less than deep sleep in most scoring algorithms. The 20 percent drop in REM doesn’t show up as a dramatic score decrease.
Late night wake events. Often captured as “restlessness” but muted in impact unless the wake events are prolonged.
HRV drop. Here’s the catch. HRV drop is measured overnight but often attributed to the sleep score generically rather than flagged as an alcohol signal specifically. The morning HRV reading might be 15 percent suppressed, but the sleep score last night can still read as reasonable.
The Result
You drink moderately. You fall asleep fast, get a bunch of deep sleep early, and the device shows you a high sleep score.
You feel mildly hungover. Your HRV is suppressed. Your recovery score the next morning is low. Your training the next day feels flat.
Two numbers from the same device. One says you slept well. The other says you didn’t recover. Both are partially right. The sleep itself was structurally worse than it looked. The recovery reflects the metabolic cost. The score chose to emphasise what looked good.
This isn’t a device malfunction. It’s a choice about what to score and how to weight it. Different devices make different choices and produce different alcohol scores.
What This Means for How to Use Sleep Scores
Three practical implications.
Don’t treat the single number as the truth. The sleep score is a weighted average of inputs, with weights that may or may not match what you care about. For alcohol specifically, the composite score underestimates cost.
Look at the components separately. If your sleep score is 84 but your overnight HRV is 20 percent below baseline and your sleeping heart rate is 8 beats above baseline, the composite score is misleading. The autonomic signals are telling you something the score is hiding.
Track a few specific metrics yourself over time. For alcohol monitoring, the two most informative metrics are overnight HRV and resting heart rate during sleep, both compared to your personal baseline. Track those separately from the sleep score. Note alcohol timing and dose. Build your own personal model of how your body responds.
What the Data Actually Looks Like After Drinking
From my own tracking, a typical response to two standard drinks finished by 8pm looks like this.
Immediate night. Sleep score 75 to 82 (often higher than baseline). Deep sleep duration 90 to 110 minutes (above normal). REM 60 to 80 minutes (below normal). Wake events 3 to 5 (slightly elevated). Sleep efficiency above normal.
Autonomic signals. Overnight HRV 15 to 25 percent below baseline. Sleeping heart rate 5 to 8 beats above baseline. Morning HRV suppressed.
Next day. Recovery score in the 40s to low 50s. Subjective energy noticeably low. Training feels flat at normal paces. Effort to pace ratio is off.
The data tells a clear story when you look at the right parts of it. Alcohol compromises recovery significantly. The sleep score alone underreports the cost.
Four standard drinks produces a much more dramatic pattern. Overnight HRV can drop 30 to 40 percent. Recovery takes 48 to 72 hours to normalise. Sleep score often still reads reasonably because the sedative effect dominates early, but by the second half of the night, wake events multiply and the scoring starts to reflect it.
The Broader Point
This is a specific example of a general issue. Composite wearable scores blend inputs with weights that reflect the designer’s choices, not objective truth about what matters for health or performance. The inputs matter. The weighting matters. Knowing what’s in the score and what’s downweighted helps you interpret it correctly.
For alcohol, the takeaway is straightforward. Don’t judge how alcohol affected your sleep by the sleep score. Check overnight HRV and sleeping heart rate. Check the next morning’s recovery score. The multi signal picture tells you the truth. The single score often doesn’t.
Green score. Destroyed legs. There are 6 blind spots in your wearable data. We wrote a free guide covering every one of them.
Green score. Destroyed legs. There are 6 blind spots in your wearable data. We wrote a free guide covering every one of them.
Download the Free Guide