The extinction of a single species is rarely an isolated event. Instead, dependent parasites, commensals, and mutualist partners (affiliates) face the risk of coextinction as their hosts or partners decline and fail. Species interactions in ecological networks can transmit the effects of primary extinctions within and between trophic levels, causing secondary extinctions and extinction cascades. Documenting coextinctions is complicated by ignorance of host specificity, limitations of historical collections, incomplete systematics of affiliate taxa, and lack of experimental studies. Host shifts may reduce the rate of coextinctions, but they are poorly understood. In the absence of better empirical records of coextinctions, statistical models estimate the rates of past and future coextinctions, and based on primary extinctions and interactions among species, network models explore extinction cascades. Models predict and historical evidence reveals that the threat of coextinction is influenced by both host and affiliate traits and is exacerbated by other threats, including habitat loss, climate change, and invasive species.