Background: Since the launch of the Safe Motherhood Initiative in 1987, the global community has called for reductions in maternal mortality in the developing world. However, reliable methods for assessing levels and trends in maternal deaths, particularly at a district level, are lacking. In increasingly-diverse countries like Indonesia, it is essential for national and local decision-makers to have timely figures to inform programmatic efforts. Aims: The aim of this PhD is to develop and pilot a community-based surveillance system (‘Surveillance by Key Informants’/SKI). The objective is a proof of principle of SKI’s reliability and feasibility, and to verify the potential for routine use at a district level in Indonesia. Methods: This PhD is comprised of two methodological studies, both designed and implemented in Serang and Pandeglang Districts of Banten Province, Indonesia, between 2004 and 2008. The Maternal Death from Informant/the Maternal Death Follow on Review (MADE-IN/MADE-FOR) method comprises retrospective, community-based surveillance of maternal deaths, involving two local informant networks. The experience from MADE-IN/MADE-FOR was used to create a prospective surveillance system (SKI), using the same networks but re-designed for routine use by district authorities. Results: MADE-IN/MADE-FOR found a high level both of maternal mortality (435 deaths per 100,000 live births) and of indirect causes (43%). The SKI findings show the considerable potential of this new approach. The two informant networks together captured about 91% of births and 92% of deaths. In general, the local stakeholders accepted SKI as a useful and realistic system for them to continuously capture local vital events. Conclusions: In a country like Indonesia where civil registration is lacking, SKI could be used to capture deaths and births in the immediate term, with MADE-IN/MADE-FOR used as a periodic census of all maternal deaths. Both approaches can provide the basis for more detailed follow-up of deaths, and so ultimately help to inform reductions in maternal mortality.