Most clustering algorithms are partitional in nature, assigning each data point to exactly one cluster. However, several real world datasets have inherently overlapping clusters in which a single data point can belong entirely to more than one cluster. This is often the case with gene microarray data since it is possible for a single gene to participate in more than one biological process. This paper deals with a novel application of consensus clustering for detecting overlapping clusters. Our approach takes advantage of the fact that results obtained by applying different clustering algorithms to the same dataset could be different and a consensus across these results could be used to detect overlapping clusters. Moreover we extend a popular model selection approach called X-means [10] to detect the inherent number of overlapping clusters in the data.