The consequences of dozing off when intending to stay awake, e.g. while driving or at work, are potentially catastrophic. The accurate assessment of this tendency is important, but is currently difficult. Several different methods give disparate results. A way out of this dilemma is suggested that involves modification of existing concepts of sleep and wakefulness to conclude the powerful influence of behaviour on sleep propensity. This propensity at a particular time depends, hypothetically, on a mutually inhibitory interaction between a sleep and a wake drive, not on the magnitude of either drive alone. Measurements of sleep propensity are partly situation-specific, whether measured objectively by laboratory tests or subjectively by a questionnaire such as the Epworth Sleepiness Scale. The latter is believed to measure a general characteristic, the average sleep propensity across a range of specified situations in daily life. Any one situational sleep propensity is not always an accurate predictor of another, even in the same subject. The Multiple Sleep Latency test should not be a gold standard for such measurements. Wider discussion and more research into “sleepiness” is needed.