The Response Time-Based Concealed Information Test (RT-CIT) is a recognition memory task used in lie detection to assess the knowledge of critical, crime-related information (Verschuere & De Houwer, 2011). Guilty participants typically negate recognition more slowly for the critical probe items than for irrelevant items. Explanations of the RT-CIT effect presume the involvement of different cognitive processes, including inhibition, task-switching, familiarity, and more conscious recollection. In the present research, we use sequential sampling models to identify the component processes in RT-CIT task performance. We compiled and reanalyzed RT-CIT data from 15 published studies (N = 1870 participants). For guilty participants, distributional analyses of probe and irrelevant item response times consistently showed delta functions with positive slopes and first-order stochastic dominance (Speckmann et al., 2007). These results imply that relatively simple assumptions about the involved cognitive processes may suffice to explain RT-CIT task performance. Next, we tested 12 plausible sequential sampling models, including versions of the Drift Diffusion Model (e.g., Ratcliff, 1978) with constant and collapsing response bounds and the Racing Diffusion Model (Tillman et al., 2020). A model with collapsing bounds fitted the data best for guilty and innocent participants and captured the observed pattern of fast errors for target items. For guilty participants, the recognition of probe items, as reflected by the drift rate, was slower than for irrelevant items. We discuss implications for the involvement of specific cognitive processes and relate the results to the dynamic recognition model (Cox & Shiffrin, 2017).