The enormous amount of biological information produced by DNA microarrays gave rise to challenges in creating more powerful analysis techniques for computer scientists. In this thesis, we examine the strengths and the weaknesses of clustering and statistical analysis and propose different integration approaches to improve the effectiveness of analyzing microarray data.The success of clustering algorithms relies on the integrity of the expression data, and they do not account for noise and non-independence among a set of experimental conditions. We used statistical analysis to take into account any dependencies and to select differentially regulated genes that were used as the inputs for clustering algorithms. On the other hand, statistical analysis relies on the integrity of the clinical data, which has some restrictions and drawbacks. We used a clustering algorithm to dynamically assign classes to the samples from the data itself and used these classes as response variables for statistical analysis.