Estimating the prevalence of the so-called “hidden populations” can be challenging, because the identification of its members is difficult due to their socially sanctionable or illegal behaviours. This article provides a critical review of the most widely used methods for estimating the size of a hard-to-reach population. All are indirect methods, based on incomplete data sources. Depending on the available data, one method can be more appropriate than another. Besides, each method must fulfil a number of requirements, and each one may be subject to specific risk of bias. To choose the most suitable method, an accurate evaluation of the available data is necessary, and. if possible several methods should be used simultaneously to be able to compare the results and to critically evaluate if these results fit with the reality.