Cognitive fatigability in multiple sclerosis represents the decrease in cognitive performance over time. It is a frequent symptom that negatively affects quality of life and ability to work. There are no objective measures of cognitive fatigability. This study aimed at quantifying cognitive fatigability despite the learning effect and to clarify whether cognitive fatigability represents a free-standing phenomenon rather than an aspect of cognitive impairment. We measured information processing speed with the Symbol Digit Modalities Test, and the number of right answers was recorded every 30 s for 180 s. We approximated the number of right answers as function of time with two logarithmic models, one including a first-order term alone and the other adding also a second-order term. The coefficient of the latter (B) may quantify performance deflection and may represent cognitive fatigability. We tested 173 patients with multiple sclerosis, including 119 cognitively impaired and 54 cognitively preserved patients, and 35 healthy subjects. The performance of cognitively preserved patients showed a deflection at the end of task that was detected neither in controls nor in cognitively impaired patients and needed a second-order term to be approximated (p < .03, F = 14.02). B was explained neither by depression nor fatigue. We proposed for the first time a method to quantify cognitive fatigue via a second-order least square fit model, easily usable in the clinical practice. By using this novel approach, cognitive fatigability results to be a free-standing phenomenon that is more evident in cognitively preserved than in cognitive impaired patients.