To cope with the uncertainty of green infrastructure planning at city scale, many cities take an adaptive approach and use learning-by-doing to improve understanding of the urban systems. However, whether that learning is worth it has been a challenge to adaptive management practitioners. In this paper, we propose an evaluation and planning framework for green infrastructure (GI) to address this issue and demonstrate its use by an application to the Wingohocking water-shed, Philadelphia, PA, USA. The framework allows users to specify possible knowledge gains from near-term actions and assess the impacts of this learning on subsequent decisions, which enables evaluation of the net benefits of alternative investment plans. In the case study, we consider two types of learning: learning to reduce uncertainty and learning to improve performance. This learning can happen through investments or knowledge transfer from experience at other locations. Estimates of cost, performance, and deterioration over time of GI and the prediction of possible knowledge gains are based on the literature and expert opinions. The results propose optimal investment strategies over a 25-year planning horizon and describe tradeoffs between the risk of poor performance and reductions in expected annual stormwater runoff. Finally, by calculating differences in expected total costs between non-adaptive, passive adaptive, and active adaptive decision-making, we quantify the economic value of learning and adaptability.