In this paper, an object-oriented (OO) framework for integrating heterogeneous database schemas for data mining is proposed. The framework is constructed on the basis of the attribute equivalence theory. The work is motivated by the observation that current data mining systems are best for knowledge discovery tasks in single databases. However, the rich sources of information provided by an increasingly large number of distributed and heterogeneous databases present an urgent need for effective and efficient mining systems that can operate on multiple databases. In our framework, OO techniques are used. The intention is to take advantages of features offered by the OO methodology. It is shown that they can be utilised effectively in the implementation.