Continuous and sustained actions in military and civilian operational environments typically lead to reduced sleep normally required to perform optimally. Because cognitive fatigue leading to defects in performance is an occupational hazard, there is a recognized need for real-time detection technologies that minimize cognitive fatigue-induced mishaps. Here, 23 individuals were subjected to 36 h of continuous wakefulness, and cognitive psychomotor vigilance and automated neuropsychological assessment metric tests were conducted over the last 24 h of wakefulness. Urine was collected prior to and during the cognitive testing period for metabolite analysis using proton NMR spectroscopy. Multivariate statistical analysis showed that temporal changes in urinary metabolite profiles mirrored cognitive performance during continuous wakefulness. Additionally, subjects identified by cognitive assessments as having a high tolerance (n=6) or low tolerance (n=6) to sleep deprivation could be classified separately with statistical confidence (p<0.001) using urinary metabolite profiles. We identified 20 specific metabolites that could be used to classify cognitive fatigue tolerance at early (0 - 12 h) and late (28 h) times during the 36-h sleep deprivation period. Many of these metabolites (11 of 20) appeared to be associated with energy metabolism or nutritional status. Analysis of subject food logs suggested that increases in dietary protein intake prior to sleep deprivation leads to improved cognitive performance. Taken together, our results indicate that urinary metabolomics may be useful for identifying metabolite markers that can be incorporated into sensor platforms to screen for cognitive performance readiness, prior to scheduling tasks requiring a high level of cognitive function.