Weeds have an important role in maintaining farmland biodiversity. This needs to be balanced with their potential negative impact on crop yield and quality. Mechanistic models of crop–weed competition are an important tool in striking this balance. A range of common UK annual weeds were screened for the eco-physiological traits required by the models. Using multivariate techniques, a number of functional groups with a similar pattern of productivity and competition were identified, based on trade-offs between traits. A scheme was developed to assign species outside of the data set to one of the groups, based on life cycle, seed mass, maximum height and time of first flowering. As well as having a similar competitive ability, species within a group also appeared to have a similar ecosystem function, in terms of supporting higher trophic groups. Two beneficial groups of species were identified that combined a relatively low competitive ability with a high importance for invertebrates and birds. The identification of functional groups in the UK arable flora is a useful tool for assessing a weed community in the context of reconciling biodiversity provision with crop production. Preserving beneficial plant functional types within the crop would complement non-cropped wildlife refuges, such as field margins.