The maternal mortality rate is a highly sensitive indicator for the health level of both women and the general population in a given geographical area. There is extensive variability among different countries, and rates are high in underdeveloped or developing areas, as in Brazil. Health authorities from various countries have launched programs aimed at reducing maternal deaths and have thus needed to estimate the actual rates to allow for a proper assessment and to control the programs' progress. However, there are many obstacles and difficulties in obtaining the real values of these measures, mainly because of incomplete data. The aim of this paper is to present some of the proposed methodologies for estimating maternal mortality rates and to call attention to the limits and biases of these methods. Based on the Brazilian case, the article also recommends an improvement in the quality and coverage of the Civil Registry, the official source of data on births and deaths.