The principle that a single large patch has higher value for biodiversity than several small patches of the same total area (SL > SS) is widely applied in conservation. We show this principle is incorrect, and that biodiversity conservation requires placing more emphasis on protection of large numbers of small patches (SS > SL). Analyzing 76 metacommunities (4401 species, 1190 patches) we demonstrate that, at equal habitat area, species richness accumulates more rapidly in large numbers of small patches twice as often as in few large patches (45.2% vs 19.9% of cases). This pattern is clear for plants, birds, mammals, and invertebrates, but reversed for herptiles. We therefore propose a new principle: for a given area, protecting the largest possible number of patches will maximize biodiversity for most taxa. Our findings indicate large untapped potential for biodiversity conservation worldwide, highlighting the need for a paradigm shift in conservation policy.