Fatigue is a frequent symptom in many clinical conditions that is still poorly understood despite having a major impact on quality of life. Here, we propose a novel approach using model-based analysis of choice behaviour to extract fatigue markers. We applied this approach to the case of low-grade glioma, with the aim of testing the hypothesis that fatigability in this condition may manifest as limited control over choice impulsivity. Patients with intact or resected glioma (n = 29) and matched healthy controls (n = 27) performed a series of behavioural tasks included in a 4 h-long neuropsychological assessment. Intertemporal choices, opposing smaller-sooner to larger-later monetary rewards, were intermixed with tasks designed to test cognitive and motor performance and to assess perceived fatigue with subjective ratings. All dependent variables were analysed with generalised linear models testing the main effects of group and time-on-task, as well as their interaction. While absent in standard measures of fatigue (subjective rating and objective performance), a significant group-by-time interaction was observed in the rate of impulsive choices: contrary to controls, patients developed a preference for the smaller-sooner option in the course of neuropsychological assessment. This preference shift was captured by computational modelling as an increase in the present bias, a parameter that assigns an additive bonus to immediate rewards. Thus, choice impulsivity was the only reliable marker that reflected the enhanced fatigability of patients relative to controls. These results suggest that the impact of glioma (or its resection) on brain functioning limits the exertion of cognitive control during decision-making. More generally, they pave the way to using model-based analysis of choice behaviour for future investigations of the many clinical conditions plagued with cognitive fatigue.