The aim of this study was to validate the Hammersmith functional motor scale for children with spinal muscular atrophy in a large cohort of 90 non-ambulant children with spinal muscular atrophy type 2 or 3. All had a baseline assessment (T0) and were reassessed either at 3 months (T1) (n = 66) or at 6 months (T2) (n = 24). Inter-observer reliability, tested on 13 children among 3 examiners, was > 95%. Of the 66 children examined after 3 months 4 had adverse effects in between assessments and were excluded from the analysis. Forty-two (68%) of the remaining 62 reassessed had no variation in scores between T0 and T1 and 13 (21%) were within +/- 1 point. 9 (37.5%) of the 24 children reassessed after 6 months had no variation in scores between T0 and T2 and another 9 (37.5%) had variations within +/- 1 point. Our study confirms previous observations of the reliability of the scale and helps to establish a baseline for assessing changes of functional ability over 3 and 6 month intervals. This information can be valuable in view of therapeutic trials.