The repetition-based truth effect refers to the phenomenon that repeated statements are more likely judged as ‘true’ than new statements. Although the truth effect has been studied for decades, it is not fully clear yet whether it reflects a reduced ability of discriminating between true and false statements or a shift in response bias towards judging statements as ‘true’. To address this question, we adopt the two-high-threshold model (2HTM) for truth judgments (Hilbig, 2012), a multinomial model assuming that two processes determine observed judgments (i.e., knowledge and guessing). The model assumes that repetition leads to a response-bias shift conditional on the lack of knowledge and is thus equivalent to the knowledge-conditional model proposed by Fazio et al. (2015). In contrast, the fluency-conditional model assumes that knowledge is used only when not relying on a general, fluency-induced response bias, which overall results in a reduced discrimination ability. To test the models against each other, we use a standard truth-effect paradigm and manipulate prior knowledge by selecting easy and difficult questions. Additionally, we manipulated guessing by informing participants about the base rate of true and false statements as well as incentivizing responses in line with the base rate manipulation. We assessed the model fit and fit the observed response patterns using receiver operating characteristic (ROC) curves.