Signal transduction pathways are very complex biochemical networks that include a large number of vital biomolecules and chemical compounds whose activities are dependent on member(s) involved in the network under consideration as well as other networks. Modularized study of such networks will be more appropriate for deciphering the underlying mechanisms that control physiological behavior of cell(s). Here we have proposed an algorithm for modularizing signal transduction network which is applied to MAPK pathway of D. melanogaster and H. sapiens. The idea "a member whose activity is controlled by maximum number of other members in the network under consideration, tends to be the center of a module" is used to divide a network into different modules. These modules are created by taking a certain level of complexity .. into account, i.e., in an expanding module, a member having more than .. relations that lie outside the module, are ignored. Recursive modularization can create modules at different stages of various complexities that finally lead to convergence of the entire network into a single module.