摘要：Summary Background Ménière's disease is characterised by severe vertigo attacks and hearing loss. Intratympanic gentamicin, the standard treatment for refractory Ménière's disease, reduces vertigo, but damages vestibular function and can worsen hearing. We aimed to assess whether intratympanic administration of the corticosteroid methylprednisolone reduces vertigo compared with gentamicin. Methods In this double-blind comparative effectiveness trial, patients aged 18–70 years with refractory unilateral Ménière's disease were enrolled at Charing Cross Hospital (London, UK) and Leicester Royal Infirmary (Leicester, UK). Patients were randomly assigned (1:1) by a block design to two intratympanic methylprednisolone (62·5 mg/mL) or gentamicin (40 mg/mL) injections given 2 weeks apart, and were followed up for 2 years. All investigators and patients were masked to treatment allocation. The primary outcome was vertigo frequency over the final 6 months (18–24 months after injection) compared with the 6 months before the first injection. Analyses were done in the intention-to-treat population, and then per protocol. This trial is registered with ClinicalTrials.gov, number NCT00802529. Findings Between June 19, 2009, and April 15, 2013, 256 patients with Ménière's disease were screened, 60 of whom were enrolled and randomly assigned: 30 to gentamicin and 30 to methylprednisolone. In the intention-to-treat analysis (ie, all 60 patients), the mean number of vertigo attacks in the final 6 months compared with the 6 months before the first injection (primary outcome) decreased from 19·9 (SD 16·7) to 2·5 (5·8) in the gentamicin group (87% reduction) and from 16·4 (12·5) to 1·6 (3·4) in the methylprednisolone group (90% reduction; mean difference −0·9, 95% CI −3·4 to 1·6). Patients whose vertigo did not improve after injection (ie, non-responders) after being assessed by an unmasked clinician were eligible for additional injections given by a masked clinician (eight patients in the gentamicin group vs 15 in the methylprednisolone group). Two non-responders switched from methylprednisolone to gentamicin. Both drugs were well tolerated with no safety concerns. Six patients reported one adverse event each: three in the gentamicin group and three in the methylprednisolone group. The most common adverse event was minor ear infections, which was experienced by one patient in the gentamicin group and two in the methylprednisolone group. Interpretation Methylprednisolone injections are a non-ablative, effective treatment for refractory Ménière's disease. The choice between methylprednisolone and gentamicin should be made based on clinical knowledge and patient circumstances. Funding Ménière's Society and National Institute for Health Research Imperial Biomedical Research Centre.