The main goal of this paper is to define a data structure to implement decision trees for classification purposes. A classification problem is defined by a set of items with defined attributes and a set of classifiers, its output is a set of classes of items. The method we followed in this paper is the known problem solving technique called augmenting data structures. The basic tree data structure is chosen. The required information to augment is defined. Also the basic operations to build, update and query this data structure are defined as algorithms. One important issue that was considered in defining this data structure is deal with ill-formed classification. Ill-formed classification may result in having an item ending in zero, one or more final classification classes. One main advantage of this data structure is that it can be used as base for software tools that facilitate the automated and interactive design, update and querying of decision trees for classification purposes.