Large publicly funded programmes of research continue to receive increased investment as interventions aiming to produce impact for the world's poorest and most marginalized populations. At this intersection of research and development, research is expected to contribute to complex processes of societal change. Embracing a co-produced view of impact as emerging along uncertain causal pathways often without predefined outcomes calls for innovation in the use of complexity-aware approaches to evaluation. The papers in this special issue present rich experiences of authors working across sectors and geographies, employing methodological innovation and navigating power as they reconcile tensions. They illustrate the challenges with (i) evaluating performance to meet accountability demands while fostering learning for adaptation; (ii) evaluating prospective theories of change while capturing emergent change; (iii) evaluating internal relational dimensions while measuring external development outcomes; (iv) evaluating across scales: from measuring local level end impact to understanding contributions to systems level change. Taken as a whole, the issue illustrates how the research for development evaluation field is maturing through the experiences of a growing and diverse group of researchers and evaluators as they shift from using narrow accountability instruments to appreciating emergent causal pathways within research for development.