Adaptive learning and teaching strategies are increasingly demanded in order to improve the efficiency and effectiveness of the education process, but few intelligent education systems exist, which are dynamic and able to satisfy individual students' requirements. In an attempt to overcome these limitations, we have developed a multi-agent education software system, which incorporates learning objects, and is based upon a learning style theory as the foundation for its adaptivity. In this article, we describe the design, implementation, and evaluation of the educational research contribution; in particular we discuss the pedagogical use of the learning objects and learning styles. We present a novel approach to the incorporation of learning style theory. The approach has been evaluated through several experiments to validate particular system functions, and the initial analysis indicates that the approach is able to handle individual students' requirements and improve the dynamic adaptivity in education systems. (Contains 1 table and 8 figures.)