Growing concern about the effects of anthropogenic sound on marine life has highlighted the need for empirical methods to study behavioral responses of marine animals to specific acoustic exposures. Some effects have been discovered by observing coincidence of effects with sound exposure, e.g. beaked whales such as Ziphius cavirostris and Mesoplodon densirostris may mass strand during sonar exercises. Sometimes new activities trigger precautionary concern, such as the potential effects of deep water seismic surveys on deep-diving endangered species, e.g. sperm whales Physeter macrocephalus. In both cases, the best way to prove that a particular sound stimulus causes a behavioral response involves experiments whereby a specific dose of sound is broadcast to an animal and the acoustic exposure and behavioral responses of the animal are measured. The present paper argues for a balance of experimental and observational studies of effects of sound on marine life, designed so that each kind of study complements the other.