Nusinersen is an antisense oligonucleotide intended for the treatment of spinal muscular atrophy. The pharmacokinetics of nusinersen, following intrathecal administrations, in the cerebrospinal fluid (CSF) and plasma of 72 pediatric patients (3 months to 17 years) with spinal muscular atrophy across 5 clinical trials was analyzed via population‐based modeling. With sparse data in the CSF and profile data in the plasma, a linear 4‐compartment model simultaneously described the time‐concentration profiles in both matrices. The typical population parameters were: Q_p = 0.572 L/h, Q_CSF = 0.069 L/h, CL_p = 2.50 L/h, CL_CSF = 0.133 L/hr, V_CSF = 0.441 L, V_p = 32.0 L, V_{systemic_tissue} = 429 L, and V_{CNS_tissue} = 258 L. A full covariate modeling approach identified baseline body weight to be a statistically and clinically relevant covariate on V_CSF, V_p, and CL_p. The model predicted that the CSF volume of distribution increased steadily with age from 0 to 2 years but became relatively steady for children >2 years. Simulations from the final model showed that age‐based dosing in children under 2 years ensured a more comparable exposure (peak concentration and area under the concentration‐time curve) across subjects in the population relative to a fixed dosing scheme. However, because no dose‐limiting toxicity has been reported in any of the trials, a fixed‐dose scheme (12 mg across all age groups) was recommended. The median terminal half‐life of nusinersen in the CSF was determined from the model to be 163 days, which supported infrequent dosing, once every 4 to 6 months in pediatric patients with spinal muscular atrophy.