Bulbs 1-6+, replaced annually with new bulbs borne terminally on rhizomes; rhizomes 1-2, secondary, short; parent bulb disappearing by anthesis except for still-functional roots and shriveled bulb coat, oblique-ovoid, 1-2 × 0.8-1.8 cm; outer coats not or only partially enclosing bulbs, brown to gray-brown, membranous, lacking cellular reticulation or cells arranged in only 2-3 rows distal to roots, ± quadrate, without fibers; inner coats white, cells very obscurely quadrate or not visible. Leaves usually deciduous with scape, withering from tip at anthesis, (1-)2, basally sheathing, sheaths not extending much above soil surface; blade solid, flat, falcate, 10-33 cm × 1.5-10 mm, margins sometimes minutely denticulate. Scape usually forming abcission layer and deciduous with leaves after seeds mature, frequently breaking at this level after pressing, solitary, ± erect, solid, flattened, winged distally, wings frequently crenulate proximal to umbel, 5-15 cm × 1-5 mm. Umbel persistent, erect, compact, 10-25-flowered, conic to hemispheric, bulbils unknown; spathe bracts persistent, 2, 8-10-veined, lanceolate, ± equal, apex acute. Flowers campanulate, 6-12 mm; tepals erect, pinkish with deeper pink midveins, lanceolate, ± equal, becoming papery in fruit, margins entire, apex acute; stamens included; anthers yellow or purple; pollen yellow; ovary obscurely crested; processes 3, central, 2-lobed, minute, margins entire; style linear, equaling stamens; stigma capitate, scarcely thickened, unlobed; pedicel 6-16 mm. Seed coat shining; cells smooth. 2n = 14. Flowering late May--Jul. Talus slopes and clay soils, including serpentine, on bald summits and ridges; 600--2500 m; B.C.; Oreg., Wash. Allium crenulatum is known only from west of the Cascade Mountains from Vancouver Island to southwestern Oregon, in Jefferson Park, Oregon, and in the Wenatchee Mountains, central Washington. The disjunct populations of Allium crenulatum in western Oregon are markedly different among themselves and from the more typical representatives to the north. It has thus far proven impossible to draw meaningful taxonomic distinctions among these populations, hence we have followed historical precedent and have placed them all in a single, highly variable species.