The stop-signal paradigm is a popular method for examining response inhibition and impulse control in psychology, cognitive neuroscience, and clinical domains because it allows the estimation of the covert latency of the stop process: the stop-signal reaction time (SSRT). In three sets of simulations, we examined to what extent SSRTs that were estimated with the popular mean and integration methods were influenced by the skew of the reaction time distribution and the gradual slowing of the response latencies. We found that the mean method consistently overestimated SSRT. The integration method tended to underestimate SSRT when response latencies gradually increased. This underestimation bias was absent when SSRTs were estimated with the integration method for smaller blocks of trials. Thus, skewing and response slowing can lead to spurious inhibitory differences. We recommend that the mean method of estimating SSRT be abandoned in favor of the integration method.