SUMMARYThe aim of this study was to identify factors other than objective sleep tendency associated with scores on the Epworth Sleepiness Scale (ESS). There were 225 subjects, of whom 40% had obstructive sleep apnoea (OSA), 16% had simple snoring, and 4.9% had snoring with sleep disruption (upper airway resistance syndrome); 9.3% had narcolepsy and 7.5% had hypersomnolence without REM sleep abnormalities; 12% had chronic fatigue syndrome; 7.5% had periodic limb movement disorder and 3% had diurnal rhythm disorders. ESS, the results of overnight polysomnography and multiple sleep latency test (MSLT) and SCL-90 as a measure of psychological symptoms were recorded. The ESS score and the mean sleep latency (MSL) were correlated (Spearman =−0.30, P<0.0001). The MSL was correlated with total sleep time (TST) and with sleep efficiency but not with apnoea/hypopnoea index. There was no association between the MSL and any aspect of SCL-90 scores, except a borderline significant association with the somatisation subscale. The ESS was correlated with TST but not with sleep efficiency or apnoea/hypopnoea index. The ESS was correlated with all subscales of the SCL-90 except psychoticism. An ESS[10 had poor sensitivity and specificity as a predictor of MSL <10 min or MSL <5 min. We conclude that the MSLT and the ESS are not interchangeable. The ESS was influenced by psychological factors by which the MSL was not affected. The ESS cannot be used to demonstrate or exclude sleepiness as it is measured by MSLT.