Eighteen males performed two vigilance tasks with static and dynamic stimuli under the influence of oxazepam (20 and 40 mg) in a placebo-controlled, double blind, crossover design. Oxazepam dose-dependently impaired overall level of performance and aggravated the decrement with time in measures of accuracy and sensitivity relative to placebo. The drug reduced the amplitudes of the P1, N1, P2N2, and P3 (dose-dependently) waves of event-related potentials (ERPs). Oxazepam aggravated the linear decline with time of the P3 amplitude only. Oxazepam impaired accuracy was related to deterioration of central processing involved in stimulus discrimination (P2N2). Impairment of response-related performance measures (RT and RI) was associated with processing manifest in the P1, N1, and P3 waves. Oxazepam effects on the amplitudes of N1 and P3 correlated with drug effects on power in alpha 1 (8-10 Hz). Drug effects on overall performance and alpha were also related; the drug effect on response speed correlated only with the drug effect on beta 1 (12.5-21 Hz). Effects of time-on-task on performance and EEG were unrelated, but oxazepam induced performance declines with time may have been caused by declines in resource allocation, as manifest in the amplitude of P3. Time effects on EEG power bands and ERP amplitudes were not significantly related to the time course of oxazepam activity. A curious dissociation emerged: both oxazepam and time-on-task impaired performance, but the drug induced a decrease of theta and alpha 1 power, whereas time-on-task increased power. Various processes play a role in performance decrements with time, and various aspects of processing may be involved in signal-detection measures which makes terms such as sensitivity quite meaningless. So-called computational processing was indistinguishable from energetic processes, which questions the validity of the distinction between these two domains. Explanations of EEG activity in terms of a unidimensional theory of arousal are untenable.