To leverage agriculture's potential to better benefit both people and nature, policymakers need clear messages about which farming practices positively impact biodiversity and yields, and when trade-offs arise. Existing reviews analyse effects of different agricultural practices on either biodiversity or yield, without considering interactions. Here, we applied multinomial and quantile regression models to synthesize global evidence of synergies and trade-offs between biodiversity and yield, using 764 paired observations (from 43 studies across 18 countries) comparing diversified and simplified farming systems. Results show that farmland diversification led to win-win outcomes for biodiversity and yield in 23% of cases, while a win for biodiversity coupled with a loss in yield was the most likely outcome (28% of cases). Yield and biodiversity responses were negatively correlated, meaning that diversifying farming systems solely in pursuit of production goals is unlikely to lead to markedly better outcomes for biodiversity, or vice-versa. Yet certain situations made win-win significantly more likely, including when crop and animal production, or multiple diversification practices (e.g., intercropping and cover crops), were combined, when no agrochemicals were applied, when diversification occurred in temperate climates, and when diversification enhanced below-ground taxa. Win-win was also more likely than lose-lose when biodiversity was measured as richness or richness-evenness, but not abundance, suggesting that in certain contexts diversified farming can effectively enhance species diversity while increasing agricultural yields. Overall, crop commodity group and bioclimatic location were amongst the most important contextual factors influencing the likelihood of a synergy or trade-off between biodiversity and yield, and diversification that accounts for these is less likely to lead to unexpected outcomes. Our novel method and up-to-date review show that farmland di