Over the past decade, diverse urban governance innovations and experiments have emerged with the declared aim to foster climate change mitigation and adaptation, involving actors at multiple levels and scales. This urban turn in environmental governance has been accompanied by normative claims and high expectations regarding a leading role of cities in coping with climate change. However, while time pressures for effective action are growing, little is known about the social learning processes involved in such urban climate governance innovations, and what they actually contribute to achieve the required transformations in urban systems. Therefore, this special issue presents eight selected papers that explore learning in urban climate governance practices in a variety of local, national and international contexts. Their findings point to a more ambiguous role of these practices as they tend to support incremental adjustments rather than deeper social learning for radical systemic change. Against this backdrop we propose a heuristic distinguishing basic modes and sources in governance learning that aims to facilitate future empirical research and comparison, thus filling a critical theory gap. Using this framework for interpretation illustrates that urban climate governance learning urgently requires more openness, parallel processes, exogenous sources, as well as novel meta-learning practices.