Eighteen subjects (ages 18-35) underwent event-related functional magnetic resonance imaging (efMRI) while performing a delayed-match-to-sample (DMS) task before and immediately after 48 h of sustained wakefulness. The DMS trial events were: a 3-s study period of either a one-, three-, or six-letter visual array; a 7-s retention interval; and a 3-s probe period, where a button press indicated whether the probe letter was in the study array. Ordinal Trend Canonical Variates Analysis (OrT CVA) was applied to the data from the probe period for trials with six-letter study lists prior to and immediately following sleep deprivation to find an activation pattern whose expression decreased with sleep deprivation in as many subjects as possible, while being present in both conditions. The first principal component of the OrT analysis identified a covariance pattern whose expression decreased as a function of sleep deprivation in 17 of 18 subjects (p<0.001). While overall expression of the pattern showed a systematic decrease with sleep deprivation, the brain regions that make up the pattern show covarying increases and decreases in activation. Regions that decreased their activation were noted in the parietal (BA 7 and 40), temporal (BA 37, 38 and 39) and occipital (BA 18 and 19) lobes; regions that increased their activation were noted in the cerebellum, basal ganglia, thalamus and the anterior cingulate gyrus (BA 32). The reduction in pattern expression with sleep deprivation for each subject was related to the change in performance on the DMS task. Subject decreases in pattern expression were correlated with reductions in recognition accuracy (p<0.05), increased intra-individual variability in reaction time (p<0.005) and increased lapsing (p<0.005).