Trees or sometimes shrubs, evergreen or deciduous, glabrous throughout. Twigs terete or angular, with lenticels, with pith or hollow. Stipules triangular, small, caducous. Leaves alternate, palmately compound; petiole long, distal portion near rachis often with glands; leaflets 3, disarticulating from petiole; petiolules short, thin when young, becoming thick in maturity; lateral leaflet blades with asymmetric base. Inflorescences at tip of new branches, corymbose racemes; rachis either with arrested growth after anthesis or growing through a leafy twig, often with obvious scars after pedicels fall; bract at base of pedicels, caducous. Flowers bisexual or unisexual by failure of one sex to develop. Pedicel long. Receptacle disklike, inner surface concave, with nectary, with sepal and petals on margin. Sepals 4, greenish, equal, obviously smaller than petals, deciduous. Petals 4, white, cream-colored, or yellow, equal, clawed, blade ovate to rhomboid with 4-6 secondary veins on each side of midvein. Stamens (8-)12-50; filaments basally connate to form a 1-4 mm androgynophore. Gynophore 2-8 cm but degenerated in staminate flower; ovary 1-locular, placentae 2, ovules many; style short or absent; stigma inconspicuous, knob-shaped. Fruit a berry, globose or ellipsoid, drooping; pericarp drying to gray, red, purple, or brown, leathery, firm, apically smooth or papillate; fruiting pedicel, receptacle, and gynophore woody and thickened. Seeds 25-50 per berry, embedded in creamy fetid or pungent mesocarp; seed coat smooth; cotyledons convolute, one longer and curved around other; radicle conical, short.

Jacobs (Blumea 12: 206. 1964) treated Crateva falcata (Loureiro) Candolle (Prodr. 1: 243. 1824; Capparis falcata Loureiro, Fl. Cochinch. 1: 331. 1790) as a doubtful species, as did FRPS (32: 490. 1999). Jacobs considered that the name applied to either C. formosensis or C. trifoliata. The fact that it was cultivated ("Habitat prope Cantonem Sinarum, inculta") suggests C. religiosa. However, Loureiro’s mention of falcate leaflets and red fruit would seem to preclude this. Jacobs could not locate a type. If a type could be found and clearly identified as either C. formosensis or C. trifoliata, then the name C. falcata would have priority in both cases. In the absence of a type and clear application of the name, it seems best to follow Jacobs and FRPS in treating C. falcata as a name of uncertain application.

About eight species: worldwide in tropics and subtropics extending north to S Japan in Asia and south to N Argentina in America; five species in China.