Abstract Background Motor imagery (MI) is increasingly used in neurorehabilitation to facilitate motor performance. Our previous study results demonstrated significantly improved walking after rhythmic-cued MI in people with multiple sclerosis (pwMS). The present feasibility study was aimed to obtain preliminary information of changes in walking, fatigue, quality of life (QoL) and MI ability following cued and non-cued MI in pwMS. The study further investigated the feasibility of a larger study and examined the reliability of a two- dimensional gait analysis system. Methods At the MS-Clinic, Department of Neurology, Medical University of Innsbruck, Austria, 15 adult pwMS (1.5-4.5 on the Expanded Disability Status Scale, 13 females) were randomised to one of three groups: 24 sessions of 17 minutes of MI with music and verbal cueing (MVMI), with music alone (MMI), or non-cued (MI). Descriptive statistics were reported for all outcomes. Primary outcomes were walking speed (Timed 25-Foot Walk) and walking distance (6-Minute Walk Test). Secondary outcomes were recruitment rate, retention, adherence, acceptability, adverse events, MI ability (Kinaesthetic and Visual Imagery Questionnaire, Time-Dependent MI test), fatigue (Modified Fatigue Impact Scale) and QoL (Multiple Sclerosis Impact Scale-29). The reliability of a gait analysis system used to assess gait synchronisation with music beat was tested. Results Participants showed adequate MI abilities. Post-intervention, improvements in walking speed, walking distance, fatigue, QoL and MI ability were observed in all groups. Success of the feasibility criteria was demonstrated by recruitment and retention rates of 8.6% (95% confidence interval, CI 5.2, 13.8%) and 100% (95% CI 76.4, 100%), which exceeded the target rates of 5.7% and 80%. Additionally, the 83% (95% CI 0.42, 0.99) adherence rate surpassed the 67% target rate. Intra-rater reliability analysis of the gait measurement instruments demonstrated excellent Intra-Class Correlation coefficients for step length of 0.978 (95% CI 0.973, 0.982) and step time of 0.880 (95% CI 0.855, 0.902). Conclusion Results from our study suggest that cued and non-cued MI are valuable interventions in pwMS who were able to imagine movements. A larger study appears feasible, however, substantial improvements to the methods are required such as stratified randomisation using a computer-generated sequence and blinding of the assessors. Trial Registration: ISRCTN ISRCTN92351899. Registered 10 December 2015.