Increasingly, scientists are being called upon to assist in the development of indicators for monitoring ecosystem health. For human health indicators, they may draw on environmental exposure, human morbidity/mortality or well-being and sustainability approaches. To improve the rigour of indicators, we propose six scientific criteria for indicator selection: (1) data availability, suitability and representativeness (of populations), (2) indicator validity (face, construct, predictive and convergent) and reliability; (3) indicator responsiveness to change; (4) indicator desegregation capability (across personal and community characteristics); (5) indicator comparability (across populations and jurisdictions); and (6) indicator representativeness (across important dimensions of concern). We comment on our current capacity to adhere to such criteria with examples of measures of environmental exposure, human health and sustainability. We recognize the considerable work still required on documenting environment-human health relationships and on monitoring potential indicators in similar ways over time. Yet we argue that such work is essential in order for science to inform policy decisions which affect the health of ecosystems and human health.