Introduction: Patients with insomnia typically have consistent physiological abnormalities including longer MSLT latencies despite poor nocturnal sleep. The present study examined demographic, subjective sleep, personality, performance, and mood correlates of MSLT scores in subjects reporting insomnia or normal sleep to identify non-EEG sleep predictors of objective sleepiness. Methods: Subjects in the insomnia group (72 males, 49 females, ages 19-52) reported a sleep problem lasting at least one year with sleep latencies of at least 45 min, or wake after sleep onset (WASO) of at least 60 min, at least four nights per week. Those in the normal group (35 males, 21 females, ages 19-53) reported no persistent sleep problem, sleep latencies no longer than 30 min, and WASO no greater than 30 min. Subjects were excluded if they reported excessive caffeine intake, substance abuse, clinical depression, or other psychiatric illness, or if laboratory recordings indicated sleep apnea, periodic leg movements, or drug-activated EEGs. Subjects were recorded polysomnographically for two nights and spent the intervening day participating in MSLT, mood, and performance tests. Metabolic rate was measured during the day. For the reported correlation and regression analyses, the mean value for the MSLT across the day was used. Results: MSLT results for the normal and insomnia groups were correlated with 44 demographic, subjective sleep, personality, psychomotor performance, and mood variables. For the normal group, significant correlations were observed between MSLT and two subjective sleep variables (number of nocturnal awakenings and amount of wake time during the night) and number of lines reviewed in a proofreading task. For the insomnia group, MSLT values were significantly correlated with the POMS Tension/Anxiety Scale, subjective nocturnal sleep latency, habitual level of caffeine consumption, and habitual tobacco use. Stepwise regression reinforced these findings. For normal sleepers, only number of nocturnal awakenings remained significantly related to MSLT with a multiple correlation of r = .41. For the insomnia group, MSLT was best predicted by a combination of caffeine use, cigarettes smoked, POMS Tension/Anxiety and a subjective sleepiness scale resulting in a multiple correlation of r = .41. Conclusions: To the extent that the MSLT is a 'pure' measure of sleepiness, it would not be expected to relate to any non-sleep variable following normal nights of sleep. The current results from normal sleepers broadly support that conclusion except for the fact that the correlation between the subjective wake variables and MSLT was positive - more frequent awakenings and wake time during the night were associated with longer MSLT values on the following day. A different pattern was observed in the insomnia group where caffeine use, tobacco use and tension/anxiety were associated with MSLT latencies. These correlations suggest either increased sensitivity to stimulants (i.e., increasing already elevated arousal), increased use of stimulants, or both. In the current data, the insomnia group smoked significantly more than normals but caffeine use did not differ between groups. Overall, the results suggest that non-sleep sources of arousal consistently influence sleep tendency in subjects reporting insomnia.