Habitat loss and degradation due to global agriculture land use is a major threat to biodiversity. Identifying agricultural management practices that mitigate these impacts is urgently needed. Thousands of experiments have been conducted worldwide in the last decades to compare the impacts of various agricultural management practices on biodiversity. The magnitudes of difference in biodiversity responses between pairs of agricultural practices, i.e. effect sizes, have now been synthesised in a growing number of meta-analyses. Yet, each meta-analysis generally focuses on a specific type of farming practice and on specific taxonomic groups, or a single region. Meta-analyses could furthermore yield different or sometimes opposite results for the similar research questions. Gathering all the effect sizes in one single dataset helps to critically assess and weigh the available evidence across all studied practices, taxonomic groups and geographical areas, and provide stakeholders a solid base to better inform their decisions. Here, we present a comprehensive dataset of 204 published meta-analyses gathering 1917 effect sizes based on more than 14 000 primary studies. We detail the effect of 9 main individual field practices (e.g. pest and disease management, amendment and fertilisation), 3 agricultural systems (e.g. organic farming, conservation agriculture) and 2 landscape level interventions (i.e. landscape complexity, land-use change). Our dataset covers numerous taxonomic groups over 14 phyla, including animals (e.g. birds, insects), microorganisms (e.g. fungi, bacteria), plants (e.g. trees, weeds). The dataset presented provides a resource to support decision-makers, farmers, and conservation ecologists alike for managing agricultural land for biodiversity