OBJECTIVE: This phase I/II study sought to explore intrathecal administration of mesenchymal stem cells (MSCs) as therapeutic approach to multiple system atrophy (MSA). METHODS: Utilizing a dose-escalation design, we delivered between 10 and 200 million adipose-derived autologous MSCs intrathecally to patients with early MSA. Patients were closely followed with clinical, laboratory, and imaging surveillance. Primary endpoints were frequency and type of adverse events; key secondary endpoint was the rate of disease progression assessed by the Unified MSA Rating Scale (UMSARS). RESULTS: Twenty-four patients received treatment. There were no attributable serious adverse events, and injections were generally well-tolerated. At the highest dose tier, 3 of 4 patients developed low back/posterior leg pain, associated with thickening/enhancement of lumbar nerve roots. Although there were no associated neurologic deficits, we decided that dose-limiting toxicity was reached. A total of 6 of 12 patients in the medium dose tier developed similar, but milder and transient discomfort. Rate of progression (UMSARS total) was markedly lower compared to a matched historical control group (0.40 ± 0.59 vs 1.44 ± 1.42 points/month, = 0.004) with an apparent dose-dependent effect. CONCLUSIONS: Intrathecal MSC administration in MSA is safe and well-tolerated but can be associated with a painful implantation response at high doses. Compelling dose-dependent efficacy signals are the basis for a planned placebo-controlled trial. CLASSIFICATION OF EVIDENCE: This phase I/II study provides Class IV evidence that for patients with early MSA, intrathecal MSC administration is safe, may result in a painful implantation response at high doses, and is associated with dose-dependent efficacy signals.