BACKGROUND: Risdiplam is an orally administered therapy that modifies pre-mRNA splicing of the survival of motor neuron 2 (SMN2) gene and is approved for the treatment of spinal muscular atrophy. The FIREFISH study is investigating the safety and efficacy of risdiplam in treated infants with type 1 spinal muscular atrophy versus historical controls. The primary endpoint of part 2 of the FIREFISH study showed that infants with type 1 spinal muscular atrophy attained the ability to sit without support for at least 5 s after 12 months of treatment. Here, we report on the safety and efficacy of risdiplam in FIREFISH part 2 over 24 months of treatment. METHODS: FIREFISH is an ongoing, multicentre, open-label, two-part study. In FIREFISH part 2, eligible infants (aged 1-7 months at enrolment, with a genetically confirmed diagnosis of spinal muscular atrophy, and two SMN2 gene copies) were enrolled in 14 hospitals in ten countries across Europe, North America, South America, and Asia. Risdiplam was orally administered once daily at 0·2 mg/kg for infants between 5 months and 2 years of age; once an infant reached 2 years of age, the dose was increased to 0·25 mg/kg. Infants younger than 5 months started at 0·04 mg/kg (infants between 1 month and 3 months old) or 0·08 mg/kg (infants between 3 months and 5 months old), and this starting dose was adjusted to 0·2 mg/kg once pharmacokinetic data were available for each infant. The primary and secondary endpoints included in the statistical hierarchy and assessed at month 12 have been reported previously. Here we present the remainder of the secondary efficacy endpoints that were included in the statistical hierarchy at month 24: the ability to sit without support for at least 30 s, to stand alone, and to walk alone, as assessed by the Bayley Scales of Infant and Toddler Development, third edition gross motor subscale. These three endpoints were compared with a performance criterion of 5% that was defined based on the natural history of type 1 spinal muscular atrophy; the results were considered statistically significant if the lower limit of the two-sided 90% CI was above the 5% threshold. FIREFISH is registered with ClinicalTrials.gov, NCT02913482. Recruitment is closed; the 36-month extension period of the study is ongoing. FINDINGS: Between March 13 and Nov 19, 2018, 41 infants were enrolled in FIREFISH part 2. After 24 months of treatment, 38 infants were ongoing in the study and 18 infants (44% [90% CI 31-58]) were able to sit without support for at least 30 s (p<0·0001 compared with the performance criterion derived from the natural history of untreated infants with type 1 spinal muscular atrophy). No infants could stand alone (0 [90% CI 0-7]) or walk alone (0 [0-7]) after 24 months of treatment. The most frequently reported adverse event was upper respiratory tract infection, in 22 infants (54%); the most common serious adverse events were pneumonia in 16 infants (39%) and respiratory distress in three infants (7%). INTERPRETATION: Treatment with risdiplam over 24 months resulted in continual improvements in motor function and achievement of developmental motor milestones. The FIREFISH open-label extension phase will provide additional evidence regarding long-term safety and efficacy of risdiplam. FUNDING: F Hoffmann-La Roche.