In this chapter, the authors advocate for an approach to corpus research that is based on explicit models and Bayesian inference. Music corpora constitute a set of naturally uncertain “observations” from which a corpus researcher wants to draw conclusions about properties that cannot be directly observed. Bayesian models make this relation explicit by defining a joint probability distribution over observed and unobserved variables that encodes the modelers’ assumptions. More broadly, listening, analysis, learning, and theory building can all be understood as inference under uncertainty from observations to unobserved causes, parameters, or entities. Bayesian modeling provides a general methodology that can be applied in each of these scenarios.