Background: The purpose of this work was to determine the efficacy of inhibiting mammalian target of rapamycin (mTOR) in pancreatic cancer preclinical models and translate preclinical observations to the clinic. Methods: Temsirolimus (20 mg Kg−1 daily) was administered to freshly generated pancreatic cancer xenografts. Tumour growth inhibition was determined after 28 days. Xenografts were characterised at baseline by gene expression and comparative genomic hybridisation. Patients with advanced, gemcitabine-resistant pancreatic cancer were treated with sirolimus (5 mg daily). The primary end point was 6-month survival rate (6mSR). Correlative studies included immunohistochemistry assessment of pathway expression in baseline tumours, drug pharmacokinetics (PKs), response assessment by FDG-PET and pharmacodynamic effects in peripheral-blood mononuclear cells (PBMCs). Results: In all, 4 of 17 xenografts (23%) responded to treatment. Sensitive tumours were characterised by gene copy number variations and overexpression of genes leading to activation of the PI3K/Akt/mTOR pathway. Activation of p70S6K correlated with drug activity in the preclinical studies. Sirolimus was well tolerated in the clinic, showed predictable PKs, exerted pathway inhibition in post-treatment PBMCs and resulted in a 6mSR of 26%. No correlation, however, was found between activated p70S6K in tumour tissues and anti-tumour effects. Conclusion: Sirolimus activity in pancreatic cancer was marginal and not predicted by the selected biomarker.