Collar et al. (1994) estimate that of the 9,672 extant species of bird, 1,111 are threatened by extinction. Here, we test whether these threatened species are simply a random sample of birds, or whether there is something about their biology that predisposes them to extinction. We ask three specific questions. First, is extinction risk randomly distributed among families? Second, which families, if any, contain more, or less, threatened species than would be expected by chance? Third, is variation between taxa in extinction risk associated with variation in either body size or fecundity? Extinction risk is not randomly distributed among families. The families which contain significantly more threatened species than expected are the parrots (Psittacidae), pheasants and allies (Phasianidae), albatrosses and allies (Procellariidae), rails (Rallidae), cranes (Gruidae), cracids (Cracidae), megapodes (Megapodidae) and pigeons (Columbidae). The only family which contains significantly fewer threatened species than expected is the woodpeckers (Picidae). Extinction risk is also not distributed randomly with respect to fecundity or body size. Once phylogeny has been controlled for, increases in extinction risk are independently associated with increases in body size and decreases in fecundity. We suggest that this is because low rates of fecundity, which evolved many tens of millions of years ago, predisposed certain lineages to extinction. Low–fecundity populations take longer to recover if they are reduced to small sizes and are, therefore, more likely to go extinct if an external force causes an increase in the rate of mortality, thereby perturbing the natural balance between fecundity and mortality.