Abstract While livestock play a vital role in supporting livelihoods of many people in Africa, they are also a major source of greenhouse gas emissions (GHG). An increasing population coupled with growing demand for livestock products means that there is an urgent need to implement cost-effective Climate Smart Agriculture (CSA) practices that can reduce emissions from livestock systems in Africa. The objective of this research was therefore to assess the effect of implementing CSA practices on milk yields and GHG emission intensities (EI) from three dairy production system types (no-graze, semi-intensive, and extensive) in Kenya. The research developed marginal abatement cost curves (MACC) to assess of the economic costs and trade-offs for the use of CSA practices in each of the dairy production systems. The research was conducted on 666 dairy farms in four counties in Kenya. Data from a farm survey and participatory workshops were used to categorise farms into production systems, estimate carbon emissions using the Agrecalc (Agricultural Resource Efficiency Calculator) tool, assess the effects of the use of CSA practices on milk yields and GHG EI, gain information on the costs of implementing CSA practices, and develop the MACCs. Our results showed that common CSA practices enhance milk yields and reduce GHG EI in dairy production systems. However, these benefits were not equally experienced by all production systems, with only clear statistical effects observed in extensive production systems. In these systems, farms using five or more CSA practices saw a 44% increase in milk production and a 25% reduction in GHG EI compared to farms not using any CSA practices. The MACCs revealed that the costs associated with the implementation of the CSA practices were higher for extensive production systems, but increased milk production meant that the net increases in value production were higher for extensive production systems. This indicates that upfront investment costs are important barriers to the use of CSA practices. Our results provide strong evidence that rural development projects are likely to be more successful when targeting farm types and using a “toolbox” approach. Moreover, the results demonstrate the importance for the establishment of policy and financing mechanisms to facilitate financing and decreasing the perceived risks involved in investing in CSA practices.