Fifty healthy male subjects were administered zolpidem (5, 10, or 20 mg), triazolam (0.5 mg) or placebo, then attempted to sleep in a non-sleep-conducive environment. Subjects were awakened at 90 min post-drug (near peak blood concentration for both drugs) and tested on several cognitive tasks, including Two Column Addition, Logical Reasoning, and a Simulated Escape Task. This was followed by a second, 3.5-h sleep period. Hypnotic efficacy of the 20 mg zolpidem (Z-20) dose was similar to that of the 0.5 mg triazolam (TRIAZ) dose, as indicated by comparably shortened sleep latencies and lengthened total sleep times. Though accuracy on most performance measures was not affected by either drug, a reduction in speed of responding on logical reasoning and addition tasks was evident for the TRIAZ group at 90 min post-drug (Ps less than 0.05). On the simulated escape task, only triazolam significantly increased the mean number of errors, and interfered with subsequent memory of the task. Thus, zolpidem had milder effects on performance than triazolam. However, 60% of the Z-20 subjects experienced mild, adverse physical reactions. Performance differences between somnogenically comparable doses of zolpidem and triazolam may be due to their differential affinities for the BZ1 and BZ2 receptor subtypes.