We propose a set of indicators to evaluate how grazing by livestock and grassland management can help preserve biodiversity in the insect communities of grassland ecosystems. These indicators can be calculated for any grassland for which management and botanical composition are known, and they do not require advanced knowledge on conservation biology and entomology. Based on the scientific literature and expert interviews, each indicator combines pressure (i.e., 16 management classes) and state variables (i.e., sward botanical composition and several of the plant functional characteristics) to assess the effects on butterfly, bumblebee, and grasshopper abundance and species richness. Each indicator was based on a multicriteria decision tree with fuzzy partitioning to account for uncertainty in the threshold values between different alternatives. The output validation of indicators was performed by comparing decision tree output with standardized measurements from 10 surveys across France, Germany, Switzerland, Wales, and Italy. The prediction of grassland suitability for butterflies was more accurate for butterfly species richness (P < 0.05) than for butterfly abundance (P = 0.10), as even under standardized transect counts, recorders will observe disproportionately more individuals when it is sunny and warm. The sensitivity of the butterfly species richness indicator was then tested using an independent dataset of 395 French grasslands. There were significant differences in predicted species richness (¿215 = 121.16; P < 0.05) among the 16 management classes, and they were ranked consistently in spite of very parsimonious assumptions in the decision tree. The prediction of grassland suitability was good for bumblebee abundance (P < 0.05) and species richness (P < 0.01). This prediction was also successful for grasshopper abundance (P < 0.05) and species richness (P = 0.05), all outliers coming from a highly diverse grassland community located on shallow soil in the