CLN2 disease is caused by deficiency in tripeptidyl peptidase-1 (TPP1), leading to neurodegeneration and death. The safety, pharmacokinetics (PK), and CNS distribution of recombinant human TPP1 (rhTPP1) were characterized following a single intracerebroventricular (ICV) or intrathecal-lumbar (IT-L) infusion to cynomolgus monkeys. Animals received 0, 5, 14, or 20 mg rhTPP1, ICV, or 14 mg IT-L, in artificial cerebrospinal fluid (aCSF) vehicle. Plasma and CSF were collected for PK analysis. Necropsies occurred at 3, 7, and 14 days post-infusion. CNS tissues were sampled for rhTPP1 distribution. TPP1 infusion was well tolerated and without effect on clinical observations or ECG. A mild increase in CSF white blood cells (WBCs) was detected transiently after ICV infusion. Isolated histological changes related to catheter placement and infusion were observed in ICV treated animals, including vehicle controls. The CSF and plasma exposure profiles were equivalent between animals that received an ICV or IT-L infusion. TPP1 levels peaked at the end of infusion, at which point the enzyme was present in plasma at 0.3% to 0.5% of CSF levels. TPP1 was detected in brain tissues with half-lives of 3–14 days. CNS distribution between ICV and IT-L administration was similar, although ICV resulted in distribution to deep brain structures including the thalamus, midbrain, and striatum. Direct CNS infusion of rhTPP1 was well tolerated with no drug related safety findings. The favorable nonclinical profile of ICV rhTPP1 supports the treatment of CLN2 by direct administration to the CNS.