Marine and coastal ecosystems provide a wide variety of recreational opportunities that are highly valued by society. For the purposes of conducting a meta-analysis we build an extensive global dataset of marine recreational ecosystem service values from the literature. Using this database we developed a number of meta-regression specifications with the objective of evaluating the study specific effects of location, ecosystem, valuation methodology and statistical estimation methods on the reported value estimates. Furthermore, the paper investigates if cultural differences between studies are an important determinant that should be considered in international (meta-analytical) value transfer. This was achieved by including a number of cultural parameters from previous societal studies and surveys into our meta-regression models. We found that accounting for differences in cultural dimensions across recreation valuation studies had a significant influence on value estimates. While a multi-level modelling approach that controls for study effects, proved to be a better fit than a standard one level specification, we found that the absolute in-sample transfer errors associated with the standard OLS model were slightly less on average based on the differences between the actual and predicted values in our meta-database.