Introduction/Background: Event-related potentials (ERPs) and time-frequency analysis are valuable tools for investigating bioelectrical activity associated with cognitive processing. The Symbol-Digit Modalities Test (SDMT) is a paper-based clinical screening and monitoring tool for cognitive processing in people with Multiple Sclerosis (pwMS). EEG may provide insights into neurophysiological processes during SDMT, potentially uncovering electrophysiological biomarkers for detecting and monitoring cognitive impairment. Our aim is to investigate the neurophysiological underpinnings of cognitive impairment in MS. Method: We developed an EEG-SDMT paradigm, displaying a symbol-number list at the top of a PC screen and 180 symbol-digit pairs (50 \% correct) in the center. Reaction times (RTs) and 32-channel EEG data were collected from 88 pwMS (57 F; age 53.70 ± 9.87; EDSS 5.97 ± 1.00) and 27 healthy controls (HC) (19 F; age 50.15 ± 12.92). ERPs and time-frequency analyses were performed. Result: SDMT-ERPs presented a positive peak in centro-parietal regions (Pz electrode) (amplitude: HC 13.14 ± 6.30 μV; pwMS 10.88 ± 6.40 μV; p = n.s.) that displayed a significantly longer latency in pwMS (606.49 ± 124.05 ms) vs HC (511.35 ± 75.55 ms) (p < 0.001). After subtracting the visual processing component from this positive peak, we identified a cognitive processing component with a significantly longer latency in pwMS (460.94 ± 123.45 ms) vs HC (370.15 ± 68.52 ms) (p < 0.01). The latencies of both peaks showed a positive correlation with EEG-SDMT RTs (r = 0.328, p < 0.001; r = 0.286, p < 0.01). PwMS showed reduced alpha (200–300 ms) and theta (300-600 ms) band increases in centro-parietal regions (p < 0.01) in comparison with HC. EEG-SDMT RTs and clinical SDMT raw scores were inversely correlated (r = −0.722; p < 0.001). Conclusion: Our EEG-SDMT paradigm provides neurophysiological correlates of SDMT performance, which was significantly affected in pwMS. The differences between pwMS and HC were primarily due to cognitive processing rather than slowed visual processing.