Acoustic communication is an integral component of social interactions in procellariid seabirds (petrels), and a substantial amount of research has been devoted to the vocalizations and vocal behavior in this family. This work has shown that petrels' calls contain information about the species, sex, and identity of the caller. Experiments have confirmed that these features are used to recognize conspecifics, mates, and other individuals in many species. Relatively little is known, however, about vocalizations in the genus Pterodroma, which contains 40% of the species in the family. My research on Pterodroma externa in the Islas Juan Fernández confirmed sexual dimorphism in the calls of this species and showed that their burrow calls differ among individuals. Both Linear Discriminant Functions and Probabilistic Neural Nets classified individuals by their calls with high accuracy. Acoustic censuses in a mixed colony of Pterodroma externa and P. longirostris showed that both of these nocturnal species increased vocal activity on nights with moonlight. Different tradeoffs between the risk of predation and the risk of collision in the dark might explain differences in the timing of their nocturnal activity. In addition, I compared aerial vocalizations in a closely related group of Pterodroma species in the subgenus Cookilaria. The similarities in the calls produced by these species suggest that vocalizations can provide useful information for understanding the phylogenetic relationships of species in this genus. Differences among the calls and activity patterns of these species, on the other hand, suggest a range of adaptations to the different environments they inhabit. One implication of these findings is that each colony of seabirds on remote islands might have an optimal pattern of activity that differs from those of colonies elsewhere. If immigrant individuals cannot adjust, these colony-specific patterns of activity could contribute to reproductive isolation between populations and thus to speciation in these birds.