AbstractNeuronal ceroid lipofuscinosis type 2 (CLN2 disease) is an ultra‐rare pediatric neurodegenerative disorder characterized by deficiency of the lysosomal enzyme tripeptidyl peptidase‐1 (TPP1). In the absence of adequate TPP1, lysosomal storage material accumulation occurs in the central nervous system (CNS) accompanied by neurodegeneration and neurological decline that culminates in childhood death. Cerliponase alfa is a recombinant human TPP1 enzyme replacement therapy administered via intracerebroventricular infusion and approved for the treatment of CLN2 disease. Here, we describe two allometric methods, calculated by scaling brain mass across species, that informed the human dose selection and exposure prediction of cerliponase alfa from preclinical studies in monkeys and a dog model of CLN2 disease: (1) scaling of dose using a human‐equivalent dose factor; and (2) scaling of compartmental pharmacokinetic (PK) model parameters. Source PK data were obtained from cerebrospinal fluid (CSF) samples from dogs and monkeys, and the human exposure predictions were confirmed with CSF data from the first‐in‐human clinical study. Nonclinical and clinical data were analyzed using noncompartmental analysis and nonlinear mixed‐effect modeling approaches. Both allometric methods produced CSF exposure predictions within twofold of the observed exposure parameters maximum plasma concentration (Cmax) and area under the curve (AUC). Furthermore, cross‐species qualification produced consistent and reasonable PK profile predictions, which supported the allometric scaling of model parameters. The challenges faced in orphan drug development place an increased importance on, and opportunity for, data translation from research and nonclinical development. Our approach to dose translation and human exposure prediction for cerliponase alfa may be applicable to other CNS administered therapies being developed.