Background: Multiple sclerosis (MS) is a chronic disease of the central nervous system, affecting approximately 2.5 million people worldwide. People with MS may experience limitations in muscular strength and endurance - including the respiratory muscles, affecting functional performance and exercise capacity. Respiratory muscle weakness can also lead to diminished performance on coughing, which may result in (aspiration) pneumonia or even acute ventilatory failure, complications that frequently cause death in MS. Training of the respiratory muscles might improve respiratory function and cough efficacy. Objectives: To assess the effects of respiratory muscle training versus any other type of training or no training for respiratory muscle function, pulmonary function and clinical outcomes in people with MS. Search methods: We searched the Trials Register of the Cochrane Multiple Sclerosis and Rare Diseases of the Central Nervous System Group (3 February 2017), which contains trials from the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, CINAHL, LILACS and the trial registry databases ClinicalTrials.gov and WHO International Clinical Trials Registry Platform. Two authors independently screened records yielded by the search, handsearched reference lists of review articles and primary studies, checked trial registers for protocols, and contacted experts in the field to identify further published or unpublished trials. Selection criteria: We included randomized controlled trials (RCTs) that investigated the efficacy of respiratory muscle training versus any control in people with MS. Data collection and analysis: One reviewer extracted study characteristics and study data from included RCTs, and two other reviewers independently cross-checked all extracted data. Two review authors independently assessed risk of bias with the Cochrane 'Risk of bias' assessment tool. When at least two RCTs provided data for the same type of outcome, we performed meta-analyses. We assessed the certainty of the evidence according to the GRADE approach. Main results: We included six RCTs, comprising 195 participants with MS. Two RCTs investigated inspiratory muscle training with a threshold device; three RCTs, expiratory muscle training with a threshold device; and one RCT, regular breathing exercises. Eighteen participants (~10%) dropped out; trials reported no serious adverse events. We pooled and analyzed data of 5 trials (N=137) for both inspiratory and expiratory muscle training, using a fixed-effect model for all but one outcome. Compared to no active control, meta-analysis showed that inspiratory muscle training resulted in no significant difference in maximal inspiratory pressure (mean difference (MD) 6.50 cmH2O, 95% confidence interval (CI) -7.39 to 20.38, P = 0.36, I2 = 0%) or maximal expiratory pressure (MD -8.22 cmH2O, 95% CI -26.20 to 9.77, P = 0.37, I2 = 0%), but there was a significant benefit on the predicted maximal inspiratory pressure (MD 20.92 cmH2O, 95% CI 6.03 to 35.81, P = 0.006, I2 = 18%). Meta-analysis with a random-effects model failed to show a significant difference in predicted maximal expiratory pressure (MD 5.86 cmH2O, 95% CI -10.63 to 22.35, P = 0.49, I2 = 55%). These studies did not report outcomes for health-related quality of life. Three RCTS compared expiratory muscle training versus no active control or sham training. Under a fixed-effect model, meta-analysis failed to show a significant difference between groups with regard to maximal expiratory pressure (MD 8.33 cmH2O, 95% CI -0.93 to 17.59, P = 0.18, I2 = 42%) or maximal inspiratory pressure (MD 3.54 cmH2O, 95% CI -5.04 to 12.12, P = 0.42, I2 = 41%). One trial assessed quality of life, finding no differences between groups. For all predetermined secondary outcomes, such as forced expiratory volume, forced vital capacity and peak flow pooling was not possible. However, two trials on inspiratory muscle training assessed fatigue using the Fatigue Severity Scale (range of scores 0-56 ), finding no difference between groups (MD, -0.28 points, 95% CI-0.95 to 0.39, P = 0.42, I2 = 0%). Due to the low number of studies included, we could not perform cumulative meta-analysis or subgroup analyses. It was not possible to perform a meta-analysis for adverse events, no serious adverse were mentioned in any of the included trials. The quality of evidence was low for all outcomes because of limitations in design and implementation as well as imprecision of results. Authors' conclusions: This review provides low-quality evidence that resistive inspiratory muscle training with a resistive threshold device is moderately effective postintervention for improving predicted maximal inspiratory pressure in people with mild to moderate MS, whereas expiratory muscle training showed no significant effects. The sustainability of the favourable effect of inspiratory muscle training is unclear, as is the impact of the observed effects on quality of life.