Ethnopharmacology relevance: Traditional herbal medicines are utilized by 27 million South Africans. Xysmalobium undulatum (Uzara) is one of the most widely used traditional medicinal plants in Southern Africa. A false belief in the safety of herbal medicine may result in liver injury. Herb-induced liver injury (HILI) range from asymptomatic elevation of liver enzymes, to cirrhosis and in certain instances even acute liver failure. Various in vitro and in vivo models are available for the pre-clinical assessment of drug and herbal hepatotoxicity. However, more reliable and readily available in vitro models are needed, which are capable of bridging the gap between existing models and real human exposure. Three-dimensional (3D) spheroid cultures offer higher physiological relevance, overcoming many of the shortcomings of traditional two-dimensional cell cultures. Aims of this study: This study investigated the hepatotoxic and anti-prolific effects of the crude X. undulatum aqueous extract during a sub-chronic study (21 days), in both a 3D HepG2/C3A spheroid model and the Sprague Dawley rat model. Methods: HepG2/C3A spheroids were treated with a known hepatotoxin, valproic acid, and crude X. undulatum aqueous extract for 21 days with continuous evaluation of cell viability and proliferation. This was done by evaluating cell spheroid growth, intracellular adenosine triphosphate (ATP) levels and extracellular adenylate kinase (AK). Sprague Dawley rats were treated with the same compounds over 21 days, with evaluation of in vivo toxicity effects on serum chemistry. Results: The results from the in vitro study clearly indicated hepatotoxic effects and possible liver damage following treatment with valproic acid, with associated growth inhibition, loss of cell viability and increased cytotoxicity as indicated by reduced intracellular ATP levels and increased AK levels. These results were supported by the increased in vivo levels of AST, ALT and LDH following treatment of the Sprague Dawley rats with valproic acid, indicative of hepatic cellular damage that may result in hepatotoxicity. The in vitro 3D spheroid model was also able to predict the potential concentration dependant hepatotoxicity of the crude X. undulatum aqueous extract. Similarly, the results obtained from the in vivo Sprague Dawley model indicated moderate hepatotoxic potential. Conclusion: The data from both the 3D spheroid model and the Sprague Dawley model were able to indicate the potential concentration dependant hepatotoxicity of the crude X. undulatum aqueous extract. The results obtained from this study also confirmed the ability of the 3D spheroid model to effectively and reliably predict the long-term outcomes of possible hepatotoxicity.