Abstract: I review literature on four different approaches to the study of traditions in animals: observation of free-living animals, laboratory experiment, armchair analysis, and field experiment. Because, by definition, a tradition entails social learning of some kind, it is difficult, perhaps impossible, to establish that a behavior is in fact traditional without knowledge of how it develops. Observations of free-living animals often provide strong circumstantial evidence of a tradition. However, even in the view of several researchers who have studied possibly traditional behaviors in natural populations, observation alone has not proven sufficient to show that social learning contributes to development of behaviors of interest. The relevance of laboratory experiments to the understanding of the development of behaviors in free-living animals is always open to challenge. Armchair analyses of field data can produce interesting hypotheses but cannot test them. Field experiments to determine how behaviors of interest develop in population members provide a promising way forward.