The California chaparral community has a rich flora of species with different mechanisms for cuing germination to postfire conditions. Heat shock triggers germination of certain species but has no stimulatory effect on a great many other postfire species that are chemically stimulated by combustion products. Previous reports have shown that charred wood will induce germination, and here we report that smoke also induces germination in these same species. Smoke is highly effective, often inducing 100% germination in deeply dormant seed populations with 0% control germination. Smoke induces germination both directly and indirectly by aqueous or gaseous transfer from soil to seeds. Neither nitrate nor ammonium ions were effective in stimulating germination of smoke-stimulated species, nor were most of the quantitatively important gases generated by biomass smoke. Nitrogen dioxide, however, was very effective at inducing germination in Caulanthus heterophyllus (Brassicaceae), Emmenanthe penduliflora (Hydrophyllaceae), Phacelia grandiflora (Hydrophyllaceae), and Silene multinervia (Caryophyllaceae). Three species, Dendromecon rigida (Papaveraceae), Dicentra chrysantha, and Trichostema lanatum (Lamiaceae), failed to germinate unless smoke treatment was coupled with prior treatment of 1 yr soil storage. Smoke-stimulated germination was found in 25 chaparral species, representing 11 families, none of which were families known for heat-shock-stimulated germination. Seeds of smoke-stimulated species have many analogous characteristics that separate them from most heat-shock-stimulated seeds, including: (1) outer seed coats that are highly textured, (2) a poorly developed outer cuticle, (3) absence of a dense palisade tissue in the seed coat, and (4) a subdermal membrane that is semipermeable, allowing water passage but blocking entry of large (molecular mass > 500) solutes. Tentative evidence suggests that permeability characteristics of this subdermal layer are altered by smoke. While the mechanism behind smoke-induced germination is not known, it appears that smoke may be involved in overcoming different blocks to germination in different species. For example, in Emmenanthe penduliflora, NO2 in smoke was sufficient to induce germination, and most forms of physical or chemical scarification also induced germination. For Romneya coulteri, NO2 alone failed to induce germination, and scarified seeds required addition of gibberellic acid. In Dicentra chrysantha, none of these treatments, nor smoke alone, induced germination, but germination was triggered by a combination of soil burial followed by smoke treatment. Smoke-stimulated species differed substantially in the duration of smoke exposure required to induce germination, and this was inversely correlated with tolerance to smoke exposure. We suggest that such differences in response may affect postfire community structure.