Funding for the reproductive healthagenda agreed at the 1994 International Conference onPopulation and Development has fallen short of estimatedrequirements. In the changed funding environment, strongerevidence on the links between reproductive health andpoverty reduction is needed. This paper reviews the evidencebase on three reproductive health outcomes: earlychildbearing, maternal survival, and unintended pregnancy.Building on the "capacities" approach to povertyassessment advocated by Amartya Sen, this evidence isorganized in a framework that includes health and educationoutcomes for mothers and children, as well as householdconsumption, and production effects. Generally speaking, theevidence on health effects is strongest, householdwell-being weakest, and education in between. Causalrelationships are difficult to establish becausereproductive health outcomes and other household-levelexplanatory variables are influenced by each other. Animportant finding is that relationships are context specificand that one cannot look at individual characteristicswithout reference to contextual variables. One reason whymuch existing research has not effectively addressed theeffects of poor reproductive health on poverty is thatstudies have relied on survey data for a single point intime. Longitudinal surveys offer greater promise. Ratherthan propose new surveys, initial research could work withexisting data resources. Research should focus on a countryor countries where 1) there has been social and economicchange over time, 2) rich-poor differences in reproductivehealth outcomes persist, and, 3) obtaining and working withpromising datasets is feasible. Pursuit of causalrelationships between poor reproductive health outcomes andpoverty should not paralyze efforts to make better use ofexisting country-level evidence in Poverty ReductionStrategies and other documents that guide resource allocation.