Classical late-infantile neuronal ceroid lipofuscinosis (LINCL), a progressive and fatal neurodegenerative disease of childhood, results from mutations in a gene (CLN2) that encodes a protein with significant sequence similarity to prokaryotic pepstatin-insensitive acid proteases. We have developed a sensitive protease activity assay that allows biochemical characterization of the CLN2 gene product in various human biological samples, including solid tissues (brain and chorionic villi), blood (buffy coat leukocytes, platelets, granulocytes, and mononuclear cells), and cultured cells (lymphoblasts, fibroblasts, and amniocytes). The enzyme has a pH optimum of 3.5 and is rapidly inactivated at neutral pH. A survey of fibroblasts and lymphoblasts demonstrated that lack of activity was associated with LINCL arising from mutations in the CLN2 gene but not other neuronal ceroid lipofuscinoses (NCLs), including the CLN6 variant LINCL, classical infantile NCL, classical juvenile NCL, and adult NCL (Kufs' disease). A study conducted using blood samples collected from classical LINCL families whose affliction was confirmed by genetic analysis indicates that the assay can distinguish homozygotes, heterozygotes, and normal controls and thus is useful for diagnosis and carrier testing. Analysis of archival specimens indicates that several specimens previously classified as LINCL have enzyme activity and thus disease is unlikely to arise from mutations in CLN2. Conversely, a specimen previously classified as juvenile NCL lacks pepinase activity and is associated with mutations in CLN2. In addition, several animals with NCL-like neurodegenerative symptoms [mutant strains of mice (nclf and mnd), English setter, border collie, and Tibetan terrier dogs, sheep, and cattle] were found to contain enzyme activity and are thus unlikely to represent models for classical LINCL. Subcellular fractionation experiments indicate that the CLN2 protein is located in lysosomes, which is consistent with its acidic pH optimum for activity and the presence of mannose 6-phosphate. Taken together, these findings indicate that LINCL represents a lysosomal storage disorder that is characterized by the absence of a specific protease activity.