The large majority of in vitro nanotoxicological studies have used immortalized cell lines for their practicality. However, results from nanoparticle toxicity testing in immortalized cell lines or primary cells have shown discrepancies, highlighting the need to extend the use of primary cells for in vitro assays. This protocol describes the isolation of mouse liver macrophages, named Kupffer cells, and their use to study nanoparticle toxicity. Kupffer cells are the most abundant macrophage population in the body and constitute part of the reticulo-endothelial system (RES), responsible for the capture of circulating nanoparticles. The Kupffer cell isolation method reported here is based on a 2-step perfusion method followed by purification on density gradient. The method, based on collagenase digestion and density centrifugation, is adapted from the original protocol developed by Smedsrød et al. designed for rat liver cell isolation and provides high yield (up to 14 x 10(6) cells per mouse) and high purity (>95%) of Kupffer cells. This isolation method does not require sophisticated or expensive equipment and therefore represents an ideal compromise between complexity and cell yield. The use of heavier mice (35-45 g) improves the yield of the isolation method but also facilitates remarkably the procedure of portal vein cannulation. The toxicity of functionalized carbon nanotubes f-CNTs was measured in this model by the modified LDH assay. This method assesses cell viability by measuring the lack of structural integrity of Kupffer cell membrane after incubation with f-CNTs. Toxicity induced by f-CNTs can be measured consistently using this assay, highlighting that isolated Kupffer cells are useful for nanoparticle toxicity testing. The overall understanding of nanotoxicology could benefit from such models, making the nanoparticle selection for clinical translation more efficient.