To date, there is no way to prove non-substance related sleepiness in a forensic context (e.g., after a traffic accident) although it is estimated that drowsy drivers cause 10-20% of all driving accidents. However, the legal situation is clear: the legislator prohibits overtired people from driving a vehicle in road traffic. An objective test for determination of non-substance related sleepiness is desirable. Metabolomics can offer a direct readout of one's phenotype. Changes in physiological functions (such as sleep-wake regulation) are reflected in endogenous metabolism and should therefore be detectable as changes in the metabolic profile, as an increase or decrease of specific metabolites correlating with hours of wakefulness. Our study aimed to assess the feasibility of discovering significant changes in the metabolome associated with hours of wakefulness. Saliva as easy-to-access samples should be used for these studies. Oral fluid samples of 13 volunteers were collected during a previous controlled sleep study that involved a sleep deprivation session and a control session in a randomized cross-over design. After a baseline night of 8 hours sleep, subjects were kept awake for 40 hours in the sleep deprivation session, whereas a 16/8 hours sleep scheme was applied during the control session. Six samples per participant were taken using Salivette® cotton swabs at defined time points across each study session during wakefulness after 0, 16, 25, 31, 38 and 40 hours, respectively. A seventh sample was taken after a recovery night that consisted of 8 hours sleep. Samples were processed using protein precipitation and dilution steps with acetonitrile and analyzed by UPLC-HRMS/MS (Sciex QTOF 6600, IDA) in an untargeted metabolomics setup. Both a reversed-phase (RP) and a hydrophilic liquid interaction (HILIC) column were used and both positive and negative ionization mode were measured to cover a wide range of metabolites. Our analyses detected numerous salivary metabolites in the presented setup (RP: 3845, HILIC: 2434), and many that underwent noticeable changes across the study time course. On the one hand, there are metabolites that increased or decreased in linear manner by hours of wakefulness during sleep deprivation but did not during control session with regular sleep regime. More importantly, some receded to baseline after recovery sleep. On the other hand, significant differences in metabolic profiles were apparent between sleep deprivation and control session. However, high inter-individual differences between study subjects impeded formulation of general statements. In addition to that, there were metabolites that followed a diurnal rhythm (i.e., circadian rhythm), like cortisol, arginine and tyrosine. We have shown that our untargeted metabolomics platform is in principle capable of analyzing salivary metabolites in sleep deprived and control subjects across a time course. Oral fluid is a rich and easily accessible matrix. Due to its simple and non-invasive sampling protocol, it is tailor-made for roadside testing and thus for forensic purposes. While rhythmically changing metabolites are not suitable to serve as marker substances for sleepiness due to their natural swings, those that show linear dynamics correlating with hours of wakefulness may serve as biomarkers. Nevertheless, identification of these remains tedious. However, several further considerations need to be addressed for a future optimized study. If not receding to baseline after recovery sleep, a metabolite cannot be related to sleepiness. High inter-individual differences come into play, too. Also, different sleep habits and therefore different sleep phenotypes are prevalent, resulting in time-shifted endogenous cycles that must be considered. Thus, a strict study protocol is needed for investigation of potential metabolic biomarkers for sleepiness to mitigate or eliminate confounding factors.