Background: Cognitive processing speed deficits are common in multiple sclerosis (MS). Despite this, the exact neural mechanism underlying MS-related slowed processing speed remains unknown. Furthermore, functional magnetic resonance imaging (fMRI) using only blood-oxygen-level-dependent (BOLD) signal may not be sensitive to MS-related metabolic changes affecting processing speed ability. Previous work has shown that cerebral metabolism in motor and visual areas are associated with performance on motor and visual tasks. However, it is unknown if task-based metabolism in the dorso-lateral prefrontal cortex (dlPFC), a region known to be involved in processing speed, is related to the slowed processing speed observed in MS. Objectives: Assess whether metabolism in a processing speed region is associated with MS-related processing speed deficits. Methods: MS and healthy control (HC) participants who met inclusion criteria were scanned using a 3T MRI scanner with a dualecho calibrated fMRI (cfMRI) sequence which provided near-simultaneous measures for both cerebral blood flow (CBF) and BOLD signal. During imaging, participants performed a block-design digit-symbol substitution task (DSST) that required the viewing of a digit-symbol pairing key and responding as to whether a probe digit-symbol pair matched the key as fast as they could using button boxes. A hypercapnia gas challenge involving periodic inhalation of room air (4 min) and 5% CO2 (6 min) permitted calculation of cerebral metabolic rate of oxygen (CMRO2). Data were preprocessed and average percent signal change (PSC) from baseline was calculated in each voxel providing BOLD and CBF time series. The anatomical region of interest (ROI) was defined as dlPFC after FreeSurfer cortical parcellation. Regression analyses were performed controlling for ROI size to assess whether BOLD, CBF, or CMRO2 could explain variability in processing speed ability. Results: An independent-samples t-test showed that the MS group had a significantly slower response time (RT) for the DSST (t=3.12, p=.003) compared to HCs. Within the MS group, regression analyses using correct-trial RTs as the dependent factor were not significant for BOLD or CBF but was significant for CMRO2 (R2=.170, p=.053) after controlling for number of voxels within the ROI. No significant regression analyses were observed within the HC group. Conclusions: Results suggest that metabolism in dlPFC explains MS-related slowed processing speed.