This chapter introduces a hierarchical modelling approach to biosecurity surveillance, arguing that this provides a common structure for representing many different existing models, ostensibly proposed within different quantitative paradigms. A Bayesian formulation is demonstrated to provide a natural framework for analyzing such hierarchical models. The chapter commences with a description of Bayesian models for estimation and prediction of pest prevalence as well as detectability, and uses this as motivation for describing the concept of Bayesian learning. The role of prior distributions in facilitating estimation with uncertainty is then discussed in detail. Attention then turns to the process of constructing hierarchical Bayesian models for surveillance, including how to model search effort, detectability, prevalence and other important features. The generality of the approach is illustrated through a commentary on stochastic scenario trees, via three-stage Bayesian hierarchical models, three-stage cluster sampling and four-stage multi-scale detection. The chapter concludes with comments on how to choose among quantitative methods, and a comparative discussion of features in the modular model-based view described here.