Extracting meaningful patterns from large databases is a relevant task in several areas of geographic research such as the interpretation of satellite images, the study of dispersion of spatial phenomena (e.g., diseases, crime), and classification of space-time behavior of individuals, to name a few. This article discusses the techniques and issues involved in spatial data mining for cluster detection and pattern recognition. The techniques range from inductive machine learning algorithms to numerical cluster detection techniques. Irrespective of the technique used, a number of issues require attention in any spatial data-mining task. These include validity testing, the selection of relevant features, interpretation of patterns, and treatment of spatial data. Approaches developed to address these issues are discussed in this article as well.