Patients with lysosomal storage diseases (LSDs) have a greatly diminished lifespan and reduced quality of life, particularly those with neurological manifestations. There are few therapeutic options available to treat the neurological signs and symptoms of LSDs. It is, therefore, imperative that efficacious and tolerable treatments are developed. Hematopoietic stem cell transplantation is carried out in some LSDs in which there is neurological involvement. However, this approach is associated with significant morbidity and mortality, and not all patients who receive this treatment exhibit improvements in cognitive signs and symptoms. A growing body of research in animal models of LSDs appears to support the efficacy of repeated delivery of recombinant lysosomal proteins via injection into the cerebrospinal fluid (CSF). Studies in dogs with mucopolysaccharidosis (MPS) Type 1 have shown that this approach enables widespread distribution of the recombinant protein within the brain, leading to a reduction in LSD pathology. Subsequent studies in MPS IIIA mice revealed that this strategy was also effective in ameliorating neuropathology and improving clinical signs in these animals. More recent studies in mice with Krabbe disease or a late infantile form of neuronal ceroid lipofuscinosis have demonstrated that delivery of recombinant proteins into the CSF may be efficacious in reducing disease pathology and neurological signs and symptoms. Whilst there are still important issues that need to be addressed, such as humoral immune responses to therapeutic protein administration and dose/frequency selection, this approach represents a medium-term option for treating these devastating conditions. This review summarizes some of the findings and challenges ahead.