A lexical decision task was used in a paradigm testing the effects of sleep loss and fatigue on performance during a 72-h period of sleep deprivation. The data were partitioned into categories of response lapses, response accuracy, and the signal detection measures of discriminability (d') and bias (β). Response lapses increased as a function of sleep loss and were fitted best by a composite equation with a major linear component and a minor rhythmic component. Response accuracy decreased as a function of sleep loss, with the rate of decrease being greater for nonwords than for words. Although d' was higher for right visual field (RVF), it decreased for both fields almost linearly as a function of sleep deprivation. The rate of decrease for RVF stimulation was greater than for left visual field (LVF) stimulation, β did not change monotonically as a function of sleep loss, but showed strong circadian rhythmicity, indicating that it was not differentially affected by sleep loss per se.