We hypothesised that HMGA1 is overexpressed in pancreatic cancers and plays a role in the malignant phenotype of pancreatic cancer cells. There are two broad components to this work: to examine the expression of HMGA1 in pancreatic cancer tissues from patients who underwent reaction and to assess if HMGA1 expression could be used as a prognostic biomarker; and to address the roles of HMGA1 in the pancreatic cancer progression, and assess the effects of therapeutic targeting of HMGA1 in experimental pancreatic cancer models. We showed that HMGA1 is overexpressed in more than 90% of pancreatic cancers and is associated with poorer postoperative survival. Tumoural HMGA1 expression was found to be an independent prognostic indicator in patients who underwent resection. Our in vitro studies suggest that HMGA1 promotes cellular invasiveness in pancreatic cancer cells through PI3-K/Akt-dependent MMP9 expression. Post-transcriptional silencing of HMGA1 suppressed metastasis in vivo. Overexpression of HMGA1 promotes PI3-K/Akt-dependent anoikis resistance and anchorage independent proliferation. The regulatory role of HMGA1 on the PI3-K/Akt pathway is novel and previously undescribed. Although HMGA1 also regulates ERK activity, the contribution of this pathway is less important in the HMGA1-induced anchorage-independent growth. In vivo silencing of HMGA1 resulted in attenuation of tumour growth in xenograft mouse model. HMGA1 represents a novel molecular determinant of chemoresistance to gemcitabine in pancreatic cancer cells. Targeted silencing of HMGA1 chemosensitised pancreatic cancer cells to gemcitabine both in vitro and in vivo. Targeting of HMGA1 may be a novel therapeutic (anti-metastatic, anti-proliferative and chemosensitising) strategy to ameliorate the aggressive phenotype of pancreatic cancer cells.