The major weakness of the current in vitro genotoxicity test systems is the inability of the indicator cells to express metabolic enzymes needed for the activation and detoxification of genotoxic compounds, which consequently can lead to misleading results. Thus, there is a significant emphasis on developing hepatic cell models, including advanced in vitro three-dimensional (3D) cell-based systems, which better imitate in vivo cell behaviour and offer more accurate and predictive data for human exposures. In this study, we developed an approach for genotoxicity testing with 21-day old spheroids formed from human hepatocellular carcinoma cells (HepG2/C3A) using the dynamic clinostat bioreactor system (CelVivo BAM/bioreactor) under controlled conditions. The spheroids were exposed to indirect-acting genotoxic compounds, polycyclic aromatic hydrocarbon [PAH; benzo(a) pyrene B(a)P], and heterocyclic aromatic amine [PhIP]) at non-cytotoxic concentrations for 24 and 96 h. The results showed that both environmental pollutants B(a)P and PhIP significantly increased the level of DNA strand breaks assessed by the comet assay. Further, the mRNA level of selected genes encoding metabolic enzymes from phase I and II, and DNA damage responsive genes was determined (qPCR). The 21-day old spheroids showed higher basal expression of genes encoding metabolic enzymes compared to monolayer culture. In spheroids, B(a)P or PhIP induced compound-specific up-regulation of genes implicated in their metabolism, and deregulation of genes implicated in DNA damage and immediate-early response. The study demonstrated that this model utilizing HepG2/C3A spheroids grown under dynamic clinostat conditions represents a very sensitive and promising in vitro model for genotoxicity and environmental studies and can thus significantly contribute to a more reliable assessment of genotoxic activities of pure chemicals, and complex environmental samples even at very low for environmental exposure relevant concentrations.