In this paper we examine a range of issues related to the analysis of health surveillance data when it is spatially-referenced. The importance of considering alarm functions derived from likelihood or Bayesian models is stressed. In addition, we focus on some new developments in predictive...