Microarray data management and analysis is emerging as an important area in bioinformatics. Due to the semi-structure nature of microarrays and the huge amount of probes in each slide, traditional database technology can not be applied directly. Microarray analyses have focused on finding differentially expressed as well as co-expressed genes among a single set of arrays associated with an experiment. We consider the problem of normalization among multiple cDNA microarray datasets from various experiments having different array platforms and non-identical sets of cDNA probes. In this paper, we provide a systematic approach for normalizing sets of arrays based on a subset of common cDNA probes shared between different datasets. Our preliminary experiments using two publicly available datasets showed that the proposed approach is feasible and can be adapted to different datasets effectively.