Legal abortions are authorized medical acts, and as such they are or can be recorded at the health facility performing them. On the other hand, the incidence of illegal, often unsafe, induced abortion is a quantity that has to be estimated. We found in the literature not less than eight methods which had been used to estimate the frequency of induced abortions: the “illegal abortion provider survey”, the "complications statistics", the “mortality statistics”, the "self-reporting", the "prospective", the "residual", the “third party reporting”, and the "expert" methods. This article details the data and methodological requirements of each of these methods, and discuss their biases. It reviews the methods' empirical record, paying special attention to the contexts in which they have been successful. It then discusses the choice of an appropriate method of estimation, given the context and the goal of the estimation.