Background: Cognitive–motor interference (CMI) has been well recognized in persons with multiple sclerosis (pwMS); however, there are limited data on effects of task difficulty. Objective: Examine (1) the effects of motor and cognitive tasks varying in difficulty on the magnitude of CMI and (2) the discriminative validity of CMI between pwMS and healthy controls (HC). Methods: Nine cognitive–motor dual-task (DT) conditions (combinations of three cognitive and three walking tasks) were examined. Outcome measures were DT-performance and dual-task cost (DTC) of gait parameters and correct answers. Task differences and overall group-effects were analysed by mixed model analysis, plus the Wilcoxon signed-rank tests or multivariate analysis of variances (MANOVAs), respectively. Results: Task effects were examined in 82 pwMS (Expanded Disability Status Scale (EDSS): 3.3 ± 1.0) and discriminative validity in a subsample (35 pwMS and 33 HC). Motor-DTC and DT-performance were affected by difficulty of both the cognitive task ( p < 0.001) and the walking condition ( p ⩽ 0.002), while cognitive-DTC only varied between cognitive tasks with a large difference in difficulty ( p ⩽ 0.005) and not between walking conditions ( p ⩾ 0.125). None of the DTCs differed between groups. Conclusion: CMI, and especially motor performance, is affected by difficulty of the DT. Although pwMS performed worse on the tasks than HC, none of the DT-conditions showed a discriminative DTC.