Introduction: In animal models of non-alcoholic fatty liver disease (NAFLD), assessment of disease severity and treatment effects of drugs rely on histopathological scoring of liver biopsies. However, little is known about the sampling variation in liver samples from animal models of NAFLD, even though several histopathological hallmarks of the disease are known to be affected by sampling variation in patients. The aim of this study was to assess the sampling variation in multiple paired liver biopsies from three commonly used diet-induced rodent models of NAFLD. Methods: Eight male C57BL/6 mice, 8 male Sprague Dawley rats and 16 female Hartley guinea pigs were fed a NAFLD-inducing high-fat diet for 16 weeks (mice and rats), 20 or 24 weeks (guinea pigs). After the initial diet period, liver sections were sampled and subsequently assessed by histopathological scoring and biochemical analyses. Results: Fibrosis was heterogeneously distributed throughout the liver in mice, manifesting as both intra- and interlobular statistically significant differences. Hepatic triglyceride content showed interlobular differences in mice, and both intra- and interlobular differences in guinea pigs (24-week time point) all of which were statistically significant. Also, hepatic cholesterol content was subject to significant intra-lobular sampling variation in mice, and hepatic glycogen content differed significantly between lobes in mice and guinea pigs. Discussion: Dependent on animal model, both histopathological and biochemical end-points differed between sampling sites in the liver. Based on these findings, we recommend that sample site location is highly standardized and properly reported in order to minimize potential sampling variation and to optimize reproducibility and meaningful comparisons of preclinical studies of NAFLD.