Opinions on how to limit the immense impact of agriculture on wild species are divided. Some think it best to retain as much wildlife as possible on farms, even at the cost of lowering yield (production per unit area). Others advocate the opposite: increasing yield so as to limit the area needed for farming, and then retaining larger areas under natural habitats. Still others support a mixture of the two extremes, or an intermediate approach. Here we summarize a model designed to resolve this disagreement, and review the empirical evidence available to date. We conclude that this evidence largely supports the second, so-called land-sparing approach to reconciling agriculture and biodiversity conservation, but that important questions remain over the generality of these findings for different biota and for ecosystem services, how best to increase yields while limiting environmental externalities, and whether there are effective, socially just, and practical mechanisms for coupling yield growth to habitat retention and restoration.