Health surveillance involves collecting public health data on chronic and infectious diseases to detect changes in disease incidence rates in order to improve public health. Timely detection of disease clusters is essential in prospective public health surveillance. Most existing health surveillance research is based on the assumption that observations from different regions are independent. This paper proposes a set of multivariate surveillance schemes generalized from well-known detection methods in multivariate statistical process control based on likelihood ratio tests. We use Monte Carlo simulations to compare these methods for health surveillance in the presence of spatial correlations. By taking advantage of correlations among regions,the proposed schemes are able to perform better than existing surveillance methods and provide faster and more accurate detection of outbreaks. An example of breast cancer in New Hampshire is presented to demonstrate the application of these methods when observations are spatially correlated counts. Copyright © 2011 John Wiley & Sons, Ltd.