Deciding the end of an epidemic is frequently associated with forthcoming changes in infectious disease control activities, including downgrading alert level in surveillance and restoring healthcare workers’ working shift back to normal. Despite the practical importance, there have been little epidemiological and laboratory methods that were proposed to determine the end of an epidemic. This short review was aimed to systematically discuss methodological principles of a small number of existing techniques and understand their advantages and disadvantages. Existing epidemiological methods have been mostly limited to a single-and-brief exposure setting, while the application to human-to-human transmissible disease epidemic with stochastic dependence structure in the observed case data has remained to be a statistical challenge. In veterinary applications, a large-scale sampling for laboratory testing has been commonly adapted to substantiate a freedom from disease, but such study has only accounted for binomial sampling process in estimating the error probability of elimination. Surveillance and mathematical modeling are two complementary instruments in the toolbox of epidemiologists. Combining their strengths would be highly beneficial to better define the end of an epidemic.