Pathogenic variants in GNPTAB and GNPTG, encoding different subunits of GlcNAc-1-phosphotransferase, cause mucolipidosis (ML) II, MLIII alpha/beta, and MLIII gamma. This study aimed to investigate the cellular and molecular bases underlying skeletal abnormalities in patients with MLII and MLIII. We analyzed bone biopsies from patients with MLIII alpha/beta or MLIII gamma by undecalcified histology and histomorphometry. The skeletal status of Gnptg and Gnptab-deficient mice was determined and complemented by biochemical analysis of primary Gnptg bone cells. The clinical relevance of the mouse data was underscored by systematic urinary collagen crosslinks quantification in patients with MLII, MLIII alpha/beta, and MLIII gamma. The analysis of iliac crest biopsies revealed that bone remodeling is impaired in patients with GNPTAB-associated MLIII alpha/beta but not with GNPTG-associated MLIII gamma. Opposed to Gnptab-deficient mice, skeletal remodeling is not affected in Gnptg mice. Most importantly, patients with variants in GNPTAB but not in GNPTG exhibited increased bone resorption. The gene-specific impact on bone remodeling in human individuals and in mice proposes distinct molecular functions of the GlcNAc-1-phosphotransferase subunits in bone cells. We therefore appeal for the necessity to classify MLIII based on genetic in addition to clinical criteria to ensure appropriate therapy.