Our increasing knowledge of the mechanisms behind the progression of pancreatic cancer (PC) has not yet translated into effective treatments. Many promising drugs have failed in the clinic, highlighting the need for better preclinical models to assess drug efficacy and characterize mechanisms of resistance. Using different experimental models, including patient-derived xenografts (PDXs), we gauged the efficacy of therapies aimed at two hallmark lesions of PCs: activation of signaling pathways by oncogenic KRAS and inactivation of tumor-suppressor genes. Although the drug targeting inactivation of tumor suppressors by DNA methylation had little effect, the inhibition of Mek, a K-Ras effector, in combination with the standard of care (chemotherapy consisting of gemcitabine/Nab-paclitaxel), reduced the growth of three out of five PC-PDXs and impaired metastasis. The two least responding PC-PDXs were composed of genetically diverse cells, which displayed sensitivities to the Mek inhibitor differing by >10-fold. Unexpectedly, our analysis of this genetic diversity unveiled different KRAS mutations. As mutation in KRAS occurs early during progression, this heterogeneity may reflect the simultaneous appearance of different malignant cellular clones or, alternatively, that cells containing two mutations of KRAS are selected during tumor evolution. In vitro and in vivo analyses indicated that the intratumoral heterogeneity, along with the selective pressure imposed by the Mek inhibitor, resulted in rapid selection of resistant cells. Together with the gemcitabine/Nab-paclitaxel backbone, Mek inhibition could be effective in treatment of PC. However, resistance because of intratumoral heterogeneity is likely to develop frequently, pointing to the necessity of identifying the factors and mechanisms of resistance to further develop this therapy.