Stimulus response latency is the delay in the onset of stimulus-evoked neuronal activity. We develop maximum likelihood and least squares estimators of stimulus response latency and present a comparison of the performance of these methods with estimators commonly used in the neuroscience literature. The formal statistical change-point estimation problem is nontrivial due to the inclusion of a 'nuisance parameter', the end of stationarity in the stimulus-evoked activity. Our results suggest that the automation of the estimation of stimulus response latency will benefit from the use of the maximum likelihood estimator.