The neuronal ceroid lipofuscinoses (NCLs) are common neurodegenerative disorders of childhood and are classified as lysosomal storage diseases since affected cells exhibit lysosomes containing ceroid and lipofuscin-like material. CLN3 is the most widely conserved NCL gene, suggesting that it has a basic eukaryotic cell function; its loss might be expected to cause the earliest onset and/or most severe disease. However, mutations in CLN3 are linked to juvenile NCL (JNCL), the latest onset and mildest form of NCL in children. We sought to explain this paradox. Almost all patients with JNCL are homozygous or heterozygous for an intragenic 1 kb deletion within CLN3, hitherto presumed to be a null mutation. We hypothesized that the 1 kb mutation may allow CLN3 residual function. We confirmed the presence of CLN3 transcripts in JNCL patient cells. When RNA silencing was used to deplete these transcripts in cells from JNCL patients, the lysosomes significantly increased in size, confirming the presence of functional protein in these cells. Consistently, overexpression of mutant CLN3 transcript caused lysosomes to decrease in size. We modelled the JNCL mutant transcripts and those corresponding to mouse models for Cln3 in Schizosaccharomyces pombe and confirmed that most transcripts retained significant function as we predicted. Therefore, we concluded that the common mutant CLN3 protein does indeed retain significant function and that JNCL is a mutation-specific disease phenotype. This finding has important consequences for recognition and diagnosis of disease caused by mutations in CLN3 and for the development of therapy for JNCL.