The plot of mortality risk against sleep duration is U-shaped: the low point sits near 7 hours, and both arms rise from there. The pattern repeats across meta-analyses — and it is often misread.
What the curve shows
The lowest relative risk of death is observed at about 7 hours of sleep per night. At 5 hours it is clearly higher; at 9 or more, higher too. You will find the full context and the chart in the Sleep silo.
What the curve does NOT show
This is observational data, not an experiment. Long sleep is far more often a symptom — of depression, inflammation, chronic disease — than their cause. Reverse causation is the main reason the right arm of the U does not mean “sleep less”.
The practical takeaway
Do not force your sleep down to an “optimal” 7 hours if you naturally need more, and do not lie in bed for 9 hours on purpose. Fix quality first: a fixed wake time and bedroom temperature do more than any single number. For concrete steps, see the sleep protocol; for a cheap lever on sleep latency, see the note on glycine.
Frequently asked questions
Does sleeping 9 hours shorten your life?
Observational data shows higher risk with long sleep, but this is most likely reverse causation: chronic disease and inflammation extend sleep. Long sleep in a healthy person is not a proven cause of death.
How much sleep is optimal?
In meta-analyses the lowest risk falls around 7 hours, but that is a population average. Your individual need may be 6.5 to 8 hours. A better signal than the number is whether you wake rested without an alarm.
Sources
- Sleep duration and all-cause mortality: a systematic review and meta-analysis (2010) · meta-analysis · PMID 20469800