Objective To retrospectively investigate safety and efficacy of nusinersen in a large cohort of adult Italian patients with spinal muscular atrophy (SMA). Methods Inclusion criteria were: (1) clinical and molecular diagnosis of SMA2 or SMA3; (2) nusinersen treatment started in adult age (>18 years); (3) clinical data available at least at baseline (T0-beginning of treatment) and 6 months (T6). Results We included 116 patients (13 SMA2 and 103 SMA3) with median age at first administration of 34 years (range 18-72). The Hammersmith Functional Rating Scale Expanded (HFMSE) in patients with SMA3 increased significantly from baseline to T6 (median change +1 point, p<0.0001), T10 (+2, p<0.0001) and T14 (+3, p<0.0001). HFMSE changes were independently significant in SMA3 sitter and walker subgroups. The Revised Upper Limb Module (RULM) in SMA3 significantly improved between T0 and T14 (median +0.5, p=0.012), with most of the benefit observed in sitters (+2, p=0.018). Conversely, patients with SMA2 had no significant changes of median HFMSE and RULM between T0 and the following time points, although a trend for improvement of RULM was observed in those with some residual baseline function. The rate of patients showing clinically meaningful improvements (as defined during clinical trials) increased from 53% to 69% from T6 to T14. Conclusions Our data provide further evidence of nusinersen safety and efficacy in adult SMA2 and SMA3, with the latter appearing to be cumulative over time. In patients with extremely advanced disease, effects on residual motor function are less clear.