Sleep deprivation impacts subjective mood states, but very little research has examined the impact on processing emotional information. In the current study, we investigated the impact of total sleep deprivation on neural responses to emotional facial expressions as well as the accuracy and speed with which these faces were categorized. Forty-nine participants completed two tasks in which they were asked to categorize emotional facial expressions as Happy, Sad, Angry, or Fearful. They were shown the 'full' expression of the emotions in one task and more subtle expressions in a second task in which expressions were 'morphed' with neutral faces so that the intensity of emotion varied. It was expected that sleep deprivation would lead to greater reactivity (indexed by larger amplitude N170 event-related potentials), particularly for negative and more subtle facial expressions. In the full face task, sleep-deprived (SD) participants were significantly less accurate than controls (C) at identifying Sad faces and slower to identify all emotional expressions. P1 was smaller and N170 was larger for the SD compared to C group, but for all emotions, indicating generalized impairment in low-level visual processing. In the more difficult morphed face task, SD participants were less accurate than C participants for Sad faces; as well, the group difference in reaction time was greatest for Sad faces. For the SD group, N170 increased in amplitude with increasing perceptual difficulty for the Fearful and Angry faces, but decreased in amplitude with increasing difficulty for Sad faces. These data illustrate that sleep deprivation led to greater neural reactivity for the threat-related negative emotions as they became more subtle; however, there was a failure to engage these perceptual resources for the processing of Sad faces. Sleep loss preferentially impacted the processing of Sad faces; this has widespread implications for sleep-deprived groups.