What Is the Epworth Sleepiness Scale and How Is It Scored?

The Epworth Sleepiness Scale is an eight-item questionnaire that asks you to rate, on a scale of zero to three, how likely you are to doze off in everyday situations like reading, watching television, or sitting in traffic. Your answers are added up to produce a score between 0 and 24, with higher numbers indicating greater daytime sleepiness. Developed in 1991 by Murray Johns at the Epworth Hospital in Melbourne, the ESS remains one of the most widely used subjective tools for measuring how sleepy a person feels during daily life. But the scale’s simplicity is both its strength and its limitation, and understanding what your score actually tells you requires knowing what the ESS can and cannot do.

What the Questionnaire Asks

The ESS presents eight scenarios and asks you to estimate your chance of dozing in each one. The situations range from passive (sitting and reading, watching TV, sitting quietly after lunch) to more active (sitting and talking to someone, sitting in a car stopped in traffic). You rate each item from 0 (“would never doze”) to 3 (“high chance of dozing”). The idea is that taken together, these eight ratings capture your general sleep propensity across the kinds of activities that fill a typical day. Johns designed the scale to measure what he called a person’s “average sleep propensity” rather than their sleepiness at a single moment in time.¹ This makes the ESS a trait measure, not a snapshot. It tells you something about your overall tendency toward sleepiness rather than how drowsy you feel right now.

That distinction matters because other tools work differently. Questionnaires like the Karolinska Sleepiness Scale ask how sleepy you are at this moment, capturing what researchers call state sleepiness.² The ESS is not designed for that. If you fill it out on a good day versus a bad day, your answers should be roughly the same, because you are reporting a general pattern, not today’s fatigue level.

How Scores Are Interpreted

The conventional cutoff for excessive daytime sleepiness is a score above 10. Most population studies and clinical guidelines use this threshold to separate “normal” from “abnormally sleepy.”³ In Johns’s original study, healthy control subjects scored well below that mark, while patients with obstructive sleep apnea, narcolepsy, and idiopathic hypersomnia scored significantly higher.⁴

The cutoff of 10 is not sacred, though. One study looking at how well the ESS identifies obstructive sleep apnea found that lowering the threshold to 8 raised sensitivity from about 66% to 76%, meaning more true cases were caught.⁵ On the other end, research examining whether a higher cutoff might improve accuracy for severe cases found that pushing the threshold up to 16 yielded 70% sensitivity but only around 56% specificity, meaning too many non-sleepy people were still being flagged.⁶ In other words, there is no single cutoff that perfectly separates sleepy from non-sleepy. The score of 10 is a reasonable middle ground for most clinical purposes, but your doctor may adjust the threshold depending on what they are screening for.

Is the ESS Reliable?

The scale holds up well on standard reliability measures. In Johns’s early validation work, a group of healthy medical students took the ESS and then took it again five months later. Their scores barely changed, and the correlation between the two rounds was strong. The questionnaire also showed high internal consistency, meaning the eight items hang together as a coherent measure of a single underlying trait.⁷

A more recent meta-analysis pooled internal-consistency data from 63 separate estimates across many studies and populations. The overall reliability score landed at about 0.82, which falls in the “good” range for a self-report questionnaire. There was, however, a lot of variation between studies, likely reflecting differences in the populations being tested and the languages the ESS was administered in.⁸ Some researchers have questioned whether the scale truly measures a single dimension of sleepiness or whether certain items cluster together. A confirmatory factor analysis in sleep apnea patients initially found that the original single-factor structure did not fit the data perfectly, though a refined version of the single-factor model ultimately did fit well.⁹ The practical takeaway is that the ESS is internally consistent enough to be clinically useful, even though it is not a perfectly clean single-construct measure.

The Gap Between Feeling Sleepy and Being Sleepy

One of the most important things to understand about the ESS is that it measures how sleepy you think you are, not how quickly you would actually fall asleep in a controlled setting. The gold-standard laboratory test for objective sleepiness is the Multiple Sleep Latency Test (MSLT), where a person lies in a dark, quiet room during the day and technicians measure how many minutes pass before sleep onset. If the ESS and the MSLT told the same story, you would expect a strong correlation between the two. They don’t.

In one early study, the ESS and MSLT showed a statistically significant but weak negative relationship. Scores of 14 or above on the ESS did tend to predict short sleep latencies on the MSLT, but below that threshold the connection was loose.¹⁰ A larger clinical study found no significant association at all. Patients who fell asleep rapidly on the MSLT and patients who stayed awake had similar ESS scores.¹¹ When close relatives or bed partners filled out the ESS on behalf of patients, the correlation with objective measures improved slightly, suggesting that self-perception is part of the problem.¹²

This disconnect is not a flaw in the ESS so much as a fundamental reality of sleep medicine. Subjective sleepiness and objective sleepiness are related but distinct things. Some people genuinely do not realize how impaired they are, and others overestimate their drowsiness because of fatigue, depression, or low energy that is not the same as the physiological drive to sleep. A low ESS score does not guarantee you are safe to drive, and a high one does not necessarily mean you will fall asleep at the wheel.

Screening for Sleep Apnea

The ESS is probably used most often in the context of obstructive sleep apnea (OSA), and here the evidence is mixed. The scale can distinguish groups of apnea patients from healthy controls, but its ability to identify individual cases is mediocre. One study using the conventional cutoff of 10 found a receiver operating characteristic (ROC) area of just 0.64 for detecting apnea, which is not much better than a coin flip.¹³ Questionnaires that incorporate anatomical and behavioral risk factors, like the STOP-Bang, consistently outperform the ESS for apnea screening. One head-to-head comparison found the STOP-Bang had sensitivity above 91% for predicting OSA, versus about 70% for the ESS.¹⁴

The reason the ESS falls short here is simple: not everyone with sleep apnea feels sleepy. Some people have significant airway obstruction overnight but never notice daytime drowsiness, or they have adapted to chronic poor sleep and no longer recognize it as abnormal. The ESS measures the symptom, not the disease. When a doctor uses it in the context of a possible apnea evaluation, a high score is a meaningful red flag, but a normal score does not rule apnea out.

Tracking Treatment Response

Where the ESS arguably shines brightest is in measuring whether a treatment is working. In patients with obstructive sleep apnea who started using CPAP (the pressurized mask worn during sleep), average ESS scores dropped from about 16 before treatment to 7 after two months, and that improvement held steady at a one-year follow-up.¹⁵ A 2024 psychometric analysis confirmed that the ESS is a valid tool for quantifying CPAP’s effect on daytime sleepiness at the group level in clinical trials.¹⁶

The ESS has also served as the primary outcome measure in drug trials. In a randomized trial of pitolisant (a wakefulness-promoting medication) for apnea patients who refused CPAP, the drug group saw ESS scores drop by about 6 points compared with roughly 4 in the placebo group, a statistically significant difference.¹⁷ The same approach has been used in trials of modafinil for narcolepsy, where patient self-assessment on the ESS improved significantly with treatment.¹⁸ Systematic reviews comparing wakefulness-promoting drugs for apnea patients routinely extract ESS change scores as one of their primary measures of benefit.¹⁹

Gender Differences and Reporting Bias

Men and women appear to answer the ESS differently in ways that do not always line up with objective measurements. In a study that compared patients’ self-scores with scores assigned by their spouses, women reported significantly higher total ESS scores than men. But the spouse data told a different story: men tended to underestimate their own sleepiness relative to what their partners observed, while women tended to overestimate theirs.²⁰ An analysis of a large online ESS dataset found that younger women scored higher than younger men, but the pattern reversed in older adults, with men in their seventies reporting more sleepiness.²¹

The clinical implications are real. In a study of narcolepsy patients, men and women reported nearly identical ESS scores (both averaging around 16), yet women demonstrated objectively worse sleepiness on the MSLT, falling asleep faster during testing than men did.²² This mismatch suggests that when men report a score of 16, they may actually be less impaired than a woman reporting the same score. Clinicians interpreting ESS results should be aware that the number alone does not mean exactly the same thing across genders.

Challenges in Older Adults

The ESS was designed around situations that assume a fairly standard adult lifestyle: riding in a car, sitting in a public place, reading after lunch. For many older adults, some of these scenarios may not apply. A person who no longer drives, rarely watches television, or does not go out in public may struggle to answer certain items. In one study of elderly patients who complained of excessive daytime sleepiness, almost 60% were unable to answer at least one ESS question. Despite all of them reporting sleepiness, only about a quarter scored above 10 on the ESS, suggesting the scale systematically underestimates sleepiness in this group. When a close relative filled out the questionnaire on the patient’s behalf, scores were significantly higher. Declining cognitive function and advancing age both widened the gap between self-reported and proxy-reported scores.²³

If you are helping an elderly parent or patient evaluate daytime sleepiness, asking a caregiver or family member to fill out the ESS in parallel can add useful information. The patient’s own score may undercount the problem, especially if cognitive decline has reduced their self-awareness of dozing episodes.

A Version for Children and Adolescents

The original ESS is written for adults, and some of its scenarios (driving, sitting in traffic) obviously do not apply to children. The ESS for Children and Adolescents (ESS-CHAD) replaces adult-specific items with age-appropriate ones while keeping the same scoring structure. Validation work using Rasch analysis found that the ESS-CHAD is a reliable and internally valid measure for adolescents between 12 and 18 years old.²⁴ Further testing in children as young as 7 with narcolepsy supported the scale’s validity in that younger, more clinical population as well.²⁵ The ESS-CHAD has been translated into multiple languages following standardized translation protocols, including Brazilian Portuguese, making it available for international use.²⁶

Sleepiness, Driving, and Real-World Risk

High ESS scores are linked to real-world consequences, most urgently to motor vehicle crashes. A prospective cohort study of over 3,200 adults found that self-reported excessive sleepiness was strongly associated with crash risk over the following year.²⁷ The same study showed that sleep apnea and short sleep duration independently predicted crashes even after adjusting for the ESS, which reinforces the point made earlier: feeling alert is not the same as being safe. People with significant sleep apnea who did not report feeling excessively sleepy still had elevated crash risk. The ESS captures one piece of the puzzle, but it cannot substitute for objective testing or common-sense awareness of fatigue.

Workplace Sleepiness and Shift Work

The ESS has been applied in occupational health research to quantify the toll of shift work and sleep deprivation. Among shift-working nurses, about 69% showed poor sleep quality based on ESS scoring.²⁸ Young doctors working night shifts scored significantly higher on the ESS than colleagues who were not acutely sleep-deprived.²⁹ In a study of medical residents, over a third exceeded the threshold for excessive daytime sleepiness, and higher ESS scores correlated with worse physical health and quality-of-life ratings.³⁰

These findings are useful because the ESS can be deployed quickly across large groups. A hospital system can screen its entire residency class in a few minutes. The limitation is that shift workers may be particularly poor at rating their own stable sleepiness trait, since their sleep schedules fluctuate so much. A resident who fills out the ESS on a recovery day after a stretch of nights may score differently than one filling it out mid-call-block. The scale was designed for people with a relatively stable routine, and shift workers live outside that assumption.

Paper Versus Digital Versions

As telemedicine and digital health platforms have expanded, researchers tested whether filling out the ESS on a screen produces the same results as the paper-and-pencil version. A secondary analysis of a randomized trial in sleep apnea patients found no significant difference between electronic and paper ESS scores. There was no evidence of bias in either direction across the full range of scores, and the level of agreement was comparable to what you see when the same person fills out the paper version twice.³¹ If your doctor sends you an online version before your appointment, you can trust that it will give the same result you would have gotten with a clipboard in the waiting room.

Sleepiness Beyond Sleep Disorders

Excessive daytime sleepiness is not exclusive to sleep apnea and narcolepsy. The ESS has been used to document elevated sleepiness in people with Parkinson’s disease. About a fifth of Parkinson’s patients in one study scored above 8, compared to just 3% of age-matched controls.³² In Parkinson’s, sleepiness and fatigue often coexist but are not the same thing. Close to half of Parkinson’s patients in one study reported excessive daytime sleepiness, and about 60% reported fatigue, with substantial overlap between the two groups.³³ The ESS is specifically aimed at sleepiness, the urge to actually fall asleep, rather than the broader sense of tiredness or low energy. If what you experience is exhaustion without the pull toward sleep, the ESS may not capture your complaint.

That distinction between sleepiness and fatigue is worth understanding for anyone filling out the ESS. Fatigue, low motivation, and physical exhaustion all make you want to lie down, but they do not all make you drift off unintentionally. The ESS asks specifically about dozing, and if your problem is more about feeling wiped out than about nodding off, your score may come back normal even though you feel terrible. Discussing both symptoms with your doctor, rather than relying on the ESS alone, gives a more complete picture.

Translated Versions Around the World

The ESS has been adapted into dozens of languages, and each validated translation goes through a standard process of forward translation, back-translation, expert review, and testing in local populations. The Korean version showed adequate reliability and validity in patients with sleep-disordered breathing.³⁴ The Arabic version was validated in healthy subjects using the same zero-to-three scoring and the same eight scenarios.³⁵ A Brazilian Portuguese adult version was tested against polysomnography results in patients with apnea, insomnia, and primary snoring, alongside healthy controls.³⁶

Cross-cultural validation matters because the ESS scenarios carry cultural assumptions. “Sitting in a car stopped for a few minutes in traffic” applies universally in car-dependent societies but may be less relevant in places where most people use public transit or rarely sit in traffic jams. “Sitting quietly after lunch without alcohol” assumes a pattern of midday meals that may or may not match local customs. Despite these concerns, the translated versions have generally held up, suggesting the underlying construct of daytime sleep propensity travels well across cultures even when specific situations feel more or less familiar.