miRNAs are small non-coding RNA that play a vital role in post-transcriptional gene regulation. They are involved in several important biological processes; hence their dysregulation has been associated with several diseases. In this study we propose a novel method to identify dysregulated miRNAs using tumor matched expression data. Applying the method to expression datasets of nine cancers from TCGA we identify dysregulated miRNAs in each of these cancers. In six cancers we see that more than 50% of the dysregulated miRNAs are up-regulated, suggesting a general trend of upregulation. We then identify transcription factors (TFs) that control the expression of dysregulated miRNAs in cancer by footprinting their upstream regions in order to build a high confidence transcriptional regulatory network contributing to the dysregulation of miRNAs. We observe that these TFs are predominantly responsible for up-regulation of miRNAs across cancers. In addition, we find that TFs that are identified in six or more cancers have different network centralities in the TF-Tf regulatory network when compared to TFs identified to contribute to dysregulation of miRNAs in a single cancer. Finally, we build cancer specific dysregulated TF-miRNA networks and identified several novel motifs including feedback loops involving TFs and miRNAs. These patterns of interactions show how TFs and miRNAs interact in a cancer specific manner and how dysregulation at one level affects the other.