Genetic Algorithm is a widely used approach in predictive data mining where data mining output can be represented by If-Then rules and discovering the best rules is done by a genetic algorithm. The main motivation for using genetic algorithms in discovery of highlevel prediction rules is that they perform a global search in the problem space and cope better with attribute interaction in compare with greedy rule induction algorithms often used in data mining (Freitas, 2001) and therefore, one can see the following papers for a wide variety of representation techniques and evolution approaches in this field: (Teng et al, 2004), (Hasanzadeh et al, 2004), (Chen & Linkens, 2004), & (Cordon et al, 1998) for evolution of weighted fuzzy rule base with simple linear genetic representation; (Golez & Dasgupta, 2002) for rule base evolution with binary tree representation; (Mendes et al, 2001) for a coevolutionary approach which evolves fuzzy rules in one process and fuzzy membership functions in another process; (Ishibuchi & Yamamoto, 2004), (de la Iglesia et al, 2003), & (Lopes et al, 1999) use multi objective optimization approaches for rule base evolution; (Ishibuchi & Yamamoto, 2002) & (Tsang et al, 2005) for two stage evolution in which one stage generates candidate rules and the other stage selects a combination of them as a final rule base; (Riquelme et al, 2003) for hierarchical representation; and some other variations in (Zhu & Guan, 2004), (Goplan et al, 2006), (Gundo et al, 2004), & (Eggermont et al, 2003). There are two basic strategies for rule base evolution task and many hybrid methods that combine the good features of these two methods. These basic approaches are Michigan approach exemplified by Holland's classifier system (Holland, 1986), and the Pittsburgh approach exemplified by Smith's LS-1 system (Smith, 1983). In this chapter, we will first study these two schools with more details in section 2 and show why there is a need for a third school, then introduce natural process of symbiogenesis in section 3 and symbiotic evolution as a novel solution for this approach in section 4. Then section 5 will present the experimental and comparison results, followed by the summary and concluding remarks in section 6.