Because lithium exerts neuroprotective effects in preclinical models of polyglutamine disorders, our objective was to assess the safety and efficacy of lithium carbonate (0.5-0.8 milliequivalents per liter) in patients with Machado-Joseph disease (spinocerebellar ataxia type 3 [MJD/SCA3]). For this phase 2, single-center, double-blind, parallel, placebo-controlled trial (ClinicalTrials.gov identifier NCT01096082), 62 patients who had MJD/SCA3 with a disease duration ≤10 years and an independent gait were randomly assigned (1:1) to receive either lithium or placebo. After 24 weeks, 169 adverse events were reported, including 50.3% in the lithium group (P = 1.00; primary safety outcome). Sixty patients (31 in the placebo group and 29 in the lithium group) were analyzed for efficacy (intention-to-treat analysis). Mean progression between groups did not differ according to scores on the Neurological Examination Score for the Assessment of Spinocerebellar Ataxia (NESSCA) after 48 weeks (-0.35; 95% confidence interval, -1.7 to 1.0; primary efficacy outcome). The lithium group exhibited minor progression on the PATA speech-rate (P = 0.002), the nondominant Click Test (P = 0.023), the Spinocerebellar Ataxia Functional Index (P = 0.003), and the Composite Cerebellar Functional Score (P = 0.029). Lithium was safe and well tolerated, but it had no effect on progression when measured using the NESSCA in patients with MJD/SCA3. This slowdown in secondary outcomes deserves further clarification.