This article evaluates four sequential sampling models—Wiener diffusion, Ornstein–Uhlenbeck (OU) diffusion, accumulator, and Poisson counter models—to understand two-choice reaction time (RT) data. Each model is fitted to response time (RT) distributions and accuracy data from three experiments, with each model augmented to account for variability in trial-to-trial differences in evidence accumulation rates, response criteria, and base RT. The Wiener diffusion model, the OU model with small-to-moderate decay, and the accumulator model with long-tailed (exponential) distributions of criteria are found to provide the best fits to the data. The article also examines the relationship between these models and three recent neurally inspired models. The evaluation involves a detailed qualitative investigation of RT and accuracy properties of the models and comparative fits to experimental data, identifying conditions under which the models make different predictions. The models are assessed using quantile probability functions (QPFs) to plot RT quantiles and probability of responses, allowing for a comprehensive examination of the joint behaviors of RT and accuracy. The study highlights the importance of variability in processing and criteria across trials in fitting the models to empirical data.This article evaluates four sequential sampling models—Wiener diffusion, Ornstein–Uhlenbeck (OU) diffusion, accumulator, and Poisson counter models—to understand two-choice reaction time (RT) data. Each model is fitted to response time (RT) distributions and accuracy data from three experiments, with each model augmented to account for variability in trial-to-trial differences in evidence accumulation rates, response criteria, and base RT. The Wiener diffusion model, the OU model with small-to-moderate decay, and the accumulator model with long-tailed (exponential) distributions of criteria are found to provide the best fits to the data. The article also examines the relationship between these models and three recent neurally inspired models. The evaluation involves a detailed qualitative investigation of RT and accuracy properties of the models and comparative fits to experimental data, identifying conditions under which the models make different predictions. The models are assessed using quantile probability functions (QPFs) to plot RT quantiles and probability of responses, allowing for a comprehensive examination of the joint behaviors of RT and accuracy. The study highlights the importance of variability in processing and criteria across trials in fitting the models to empirical data.