Moment-to-moment variations in hearing and speech perception have long been observed. Depending on the researcher's theoretical position, the observed fluctuations have been attributed to measurement error or to internal, nonsensory factors such as fluctuations in attention. While cognitive performance has been shown to fluctuate from day to day over longer time, such fluctuations have not been quantified for speech perception, despite being well-recognized by clinical audiologists and hearing-impaired patients. In three studies, we aimed to explore and quantify the magnitude of daily variability in speech perception and to investigate whether such variability goes beyond test unreliability. We also asked whether intraindividual variability depends on overall speech perception performance as observed in different groups of individuals. Older adults with objective hearing impairment and mostly hearing aids (N₁ = 45), with subjective hearing problems but no hearing aids (N₂ = 113), and younger adults without hearing problems (N₃ = 20) participated in three ecological momentary assessment studies. They performed a digit-in-noise test two to three times a day for several weeks. Variance heterogeneous linear mixed-effects models indicated reliable intraindividual variability in speech perception and substantial individual differences in daily variability. A protective factor against daily fluctuations is a higher average speech perception. These studies show that day-to-day variations in speech perception cannot simply be attributed to test unreliability and pave the way for investigating how psychological states that do not vary from moment-to-moment, but rather from day to day, predict variations in speech perception. (PsycInfo Database Record (c) 2023 APA, all rights reserved).