Field margins are key features of agricultural landscapes, especially in northwestern Europe where field margins planted with trees, i.e. hedgerows, are the dominant type of semi-natural habitats (Burel and Baudry, 1995). They have a wide range of purposes, including agronomical, ecological and cultural functions (Baudry et al., 2000). In particular, they support farmland biodiversity such as carabids by providing overwintering sites, food resources and shelters (Meek et al., 2002). However, little is known about the effects of management practices of field margins on biodiversity, and their historical dimension, in terms of past management practices and landscape context, is rarely considered. Here, we assessed the relative influence of current and past management practices and landscape heterogeneity (during the last five years) on the local diversity (species richness and composition) of carabid assemblages of field margins in agricultural landscapes of Brittany, France. More specifically, we tested whether there is a time lag in carabid diversity response to local management and landscape factors. We found that current carabid diversity was best explained by management practices and landscape heterogeneity measured four or five years ago. This result suggests the existence of a time lag in the response of carabid assemblages to past environmental conditions of field margins. The influence of past management practices was overall higher than the influence of past landscape heterogeneity. Carabid diversity was higher in grazed or sprayed field margins probably due to increased heterogeneity in habitat conditions. Field margins surrounded by grasslands and crops harbored carabid species associated to open habitats whilst forest carabid species dominated field margins surrounded by woodland. Landscape effect was scale-dependent and mostly detected at fine spatial scale, within 50 m around field margins. Our study highlights the importance of taking account of time to understand current patterns of carabid species. We suggest that landscape management actions and conservation strategies consider the possibility of time-lagged responses when studying biodiversity patterns in agricultural landscapes.