Cities increasingly address climate change, e.g. by pledging city-level emission reduction targets. This is puzzling for the provision of a global public good: what are city governments’ reasons for doing so, and do pledges actually translate into emission reductions? Empirical studies have found a set of common factors which relate to these questions, but also mixed evidence. What is still pending is a theoretical framework to explain those findings and gaps. This paper thus develops a theoretical public choice model. It features economies of scale and distinguishes urban reduction targets from actual emission reductions. The model is able to explain the presence of targets and public good provision, yet only under specified conditions. It is also able to support some stylized facts from the empirical literature, e.g. on the effect of city size, and resolves some mixed evidence as special cases. Larger cities chose more ambitious targets if marginal net benefits of mitigation rise with city size—if they set targets at all. Whether target setting is more likely for larger cities depends on the city type. Two types are obtained. The first type reduces more emissions than a free-riding city. Those cities are more likely to set a target when they are larger. However, they miss the self-chosen target. Cities of the second type reach their target, but mitigate less than a free-riding city. A third type does not exist. With its special cases, the model can thus guide further empirical and theoretical work.