This investigation examines how the sleep propensity (SP) in one test situation, such as the Multiple Sleep Latency Test (MSLT), is related to sleepiness in daily life, as assessed by the Epworth Sleepiness Scale (ESS). This is a self-administered questionnaire, the item scores from which provide a new method for measuring SPs in eight different real-life situations. The ESS item scores were analyzed separately in four groups of subjects: 150 adult patients with a variety of sleep disorders, 87 medical students who answered the ESS on two occasions 5 months apart, 44 patients who also had MSLTs and 50 patients whose spouses also answered the ESS about their partner's sleepiness. The ESS item scores were shown to be reliable (mean rho = 0.56, p < 0.001). The SP measured by the MSLT was related to three of the eight item scores in a multiple regression (r = 0.64, p < 0.001). The results of nonparametric ANOVA, Spearman correlations, Wilcoxon's t tests, item and factor analysis suggest that individual measurements of SP involve three components of variation in addition to short-term changes over periods of hours or days: a general characteristic of the subject (his average SP), a general characteristic of the situation in which SP is measured (its soporific nature) and a third component that is specific for both subject and situation. The SP in one test situation, including the MSLT, may not be a reliable indicator of a subject's average SP in daily life. Perhaps we should reexamine the current concept of daytime sleepiness and its measurement.