To test the hypothesis that fatigue-induced performance decline in surgical residents is associated with changes in brain function as detected by functional near-infrared spectroscopy. Surgical residents (n = 7) participated in a prospective study involving 2-hourly objective measurements of neurocognitive skill (arithmetic calculations using Nintendo "brain training"), technical performance (surgical knot tying on a trainer, and monitoring time taken, path length and number of movements), and introspective fatigue (questionnaire-based) across 10 hours of acute sleep deprivation (10:00 PM to 8:00 PM. Simultaneously, changes in cortical oxyhemoglobin (HbO₂), deoxyhemoglobin (HHb), and total hemoglobin (HbT), inferring prefrontal function, were recorded by using functional near-infrared spectroscopy. Arithmetic performance remained stable despite increasing levels of subject fatigue (time: P = 0.07, errors: P = 0.70, efficiency: P = 0.58). Technical skill improved between the first (10:00 PM and the second (12:00 AM sessions (P < 0.05) and stabilized thereafter (12:00 AM to 8:00 AM. Greater activation was required to complete cognitive versus technical drills. Stimulus type (0: cognitive, 1: technical) was found to be an independent predictor of changes in cortical excitation (HbO₂: P < 0.01, HHb: P < 0.05, HbT: P < 0.01). Cortical responses to the cognitive task increased over the course of the simulated night shift. In addition, "time interval" was observed to be an independent predictor of cortical hemodynamic change (HbO₂: P < 0.01, HbT: P < 0.01). Neurocognitive tasks may tax the sleep-deprived resident more than well-learned technical skills. Performing cognitive skills at night, such as decision making, may depend upon enhanced prefrontal recruitment indicative of a focused attentional strategy and/or compensation to sleep deprivation. Further work should focus on determining whether errors in performance are associated with attentional lapses and failure of cortical compensation.