In the Stroop task, color words are presented in colored fonts and the task of the subject is to either name the word or name the color. If the word and font color are in agreement, then the stimulus is said to be congruent (e.g., RED in red font color); however, if the word and font color are not in agreement, the stimulus is said to be incongruent (e.g., RED in blue font color). Conflict in the Stroop task is measured by both RT and accuracy. In prior research, the amount of conflict differs depending on the response modality, vocal versus manual. We applied a model for multichoice decision-making (and confidence), the RTCON2 model (Ratcliff & Starns, 2013) to the data from 4 experiments, 2 with 2-choice manual responses, 1 with 4-choice manual touch screen responses, 1 with both 4-choice vocal responses, and 4-choice manual keyboard responses. Changes in the rate of information accumulation captured conflict effects for the manual-response versions, but not for the vocal-response version. Adding an extra nondecision time parameter allowed RTCON2 to account for the data patterns in the vocal-response version. However, to fully understand conflict in the vocal-response Stroop task, a model of conflict processing in the vocal word production system must be developed that would explain the additional processing time in the nondecision time parameter.