The playback of sounds to animals to assess their behavioural responses presents a powerful tool to study animal cognition in the wild. While playbacks are commonly used to study acoustic responses in birds and other terrestrial animals, their application to the study of marine mammal cognition so far has been limited. A survey of the published literature on field playback experiments with marine mammals identified 46 studies, with a trend towards increased use of playback approaches in recent years. Field playbacks to marine mammals have been used to address questions of wildlife management, the impact of anthropogenic noise, acoustic interactions between predators and prey, individual and kin recognition, as well as the function of communicative sounds. This paper summarizes the major findings of marine mammal playbacks to date and reviews recent advances in the design and execution of playback experiments, with special reference to marine mammals. Issues concerning appropriate presentation of acoustic stimuli, appropriate quantification of behavioural responses, as well as appropriate control and replication of treatments are discussed. An analysis of replication in marine mammal playbacks showed that the use of a small number of playback stimuli to conduct multiple playback trials (pseudoreplication) was common. This overview of playback experiments in the study of marine mammal cognition in the wild showed that such approaches contribute significantly to the field; however, in many cases, there appears to be substantial room for improvement of playback procedure and experimental design.