Previous studies have shown a correspondence between the abundance of particular plant species and methane flux. Here, we apply multivariate analyses, and weighted averaging, to assess the suitability of vegetation composition as a predictor of methane flux. We developed a functional classification of the vegetation, in terms of a number of plant traits expected to influence methane production and transport, and compared this with a purely taxonomic classification at species level and higher. We applied weighted averaging and indirect and direct ordination approaches to six sites in the United Kingdom, and found good relationships between methane flux and vegetation composition (classified both taxonomically and functionally). Plant species and functional groups also showed meaningful responses to management and experimental treatments. In addition to the United Kingdom, we applied the functional group classification across different geographical regions (Canada and the Netherlands) to assess the generality of the method. Again, the relationship appeared good at the site level, suggesting some general applicability of the functional classification. The method seems to have the potential for incorporation into large-scale (national) greenhouse gas accounting programmes (in relation to peatland condition/management) using vegetation mapping schemes. The results presented here strongly suggest that robust predictive models can be derived using plant species data (for use in national-scale studies). For trans-national-scale studies, where the taxonomic assemblage of vegetation differs widely between study sites, a functional classification of plant species data provides an appropriate basis for predictive models of methane flux.