Significant attention has been directed to evaluations of early childhood education and Head Start. These evaluations have focused, primarily, on finding what works. The more useful research question going forward might be, "What works better?" Or, perhaps, "What works better for whom?" Those who care about program impact, including those who run the programs and those who fund them, need researchers to help them learn how to adjust their practices to get greater impact, including greater impact for every dollar spent. This requires fine-tuning and learning over time, what might be called iterative experiments or measured trials. This is more likely to happen if we can create a learning system with researchers working more closely with delivery agents to test variations of practice to find increasingly effective and cost-effective early childhood practices. This session will explore what is already happening in this area and work to identify research questions as well as willing academics and practitioner partners. It will also explore how to create researcher/practitioner partnerships that work well.