Here we develop a monthly gridded dataset of northern (> 45 N) wetland methane (CH₄) emissions. The data product is derived using a random forest machine-learning technique and eddy covariance CH₄ fluxes from 25 wetland sites. Annual CH₄ emissions from these wetlands calculated from the derived data product are comparable to prior studies focusing on these areas. This product is an independent estimate of northern wetland CH₄ emissions and hence could be used, e.g. for process model evaluation.