Aim. Fruit colours attract animal seed dispersers, yet the causes of fruit colour diversity remain controversial. The lack of knowledge of large-scale spatial patterns in fruit colours has limited our ability to formulate and test alternative hypotheses to explain fruit colour, fruit size, and fruit colour diversity. We describe spatial (especially latitudinal) variation in fruit colour, colour diversity, and length, and test for correlations between fruit colour, length, and plant habit. Location. Global. Time period. Present-day. Major taxa studied. Seed plants. Methods. We assembled a database of fruit traits for 13,178 fleshy-fruited plant species spanning 136 sites around the world. To assess whether fruit colour categories correspond with spectral reflectances, we tested for clustering of hue, chroma, and saturation for 236 species for which we had reflectance data. We then quantified latitudinal gradients in fruit colour, fruit length, and fruit colour diversity while controlling for the effects of plant habit, and whether colour categories varied with respect to average fruit size. Results. Colour categories corresponded well with reflectance data. The tropics show high colour diversity, while red fruits progressively constitute a higher proportion of the fleshy-fruited plant community towards high latitudes. All mammal-associated colours (green, orange, brown, and yellow) are more common in the tropics than at high latitudes. Fruit length also increases towards the tropics. Main conclusions. Tropical communities tend to have diverse fruit colours, including many mammal-associated fruit colours, while high latitude communities contain a higher percentage of red-fruited species. The correlation between colour and size is strong, and some latitudinal patterns may be partly driven by changes in fruit size. Differences in geography and in the history of plant lineages in the southern versus the northern hemisphere may help to explain some biogeographic patterns, but alternative hypotheses related to fruit defence, development, and metabolic costs are plausible.