3-Nitrobenzanthrone (3-NBA) is a byproduct of diesel exhaust and is highly present in industrial and populated areas. Inhalation of 3-NBA results in formation of N-(2'-deoxyguanosin-8-yl)-3-aminobenzanthrone (dGC8-N-ABA), a bulky DNA lesion that is of concern due to its mutagenic and carcinogenic potential. If dGC8-N-ABA is not bypassed during genomic replication, the lesion can stall cellular DNA replication machinery, leading to senescence or apoptosis. We have previously used running start assays to demonstrate that human DNA polymerases eta (hPolη) and kappa (hPolκ) are able to catalyze translesion DNA synthesis (TLS) across a site-specifically placed dGC8-N-ABA in a DNA template. Consistently, gene knockdown of hPolη and hPolκ in HEK293T cells reduces the efficiency of TLS across dGC8-N-ABA by ∼25 and ∼30%, respectively. Here, we kinetically investigated why hPolκ paused when bypassing and extending from dGC8-N-ABA. Our kinetic data show that correct dCTP incorporation efficiency of hPolκ dropped by 116-fold when opposite dGC8-N-ABA relative to undamaged dG, leading to hPolκ pausing at the lesion site observed in the running start assays. The already low nucleotide incorporation fidelity of hPolκ was further decreased by 10-fold during lesion bypass, and thus, incorrect nucleotides, especially dATP, were incorporated opposite dGC8-N-ABA with comparable efficiencies as correct dCTP. With regard to the dGC8-N-ABA bypass product extension step, hPolκ incorporated correct dGTP onto the damaged DNA substrate with a 786-fold lower efficiency than onto the corresponding undamaged DNA substrate, which resulted in hPolκ pausing at the site in the running start assays. Furthermore, hPolκ extended the primer-terminal matched base pair dC:dGC8-N-ABA with a 100-1000-fold lower fidelity than it extended the undamaged dC:dG base pair. Together, our kinetic results strongly indicate that hPolκ was error-prone during TLS of dGC8-N-ABA.