Inhibition of the HSP90 chaperone results in depletion of many signaling proteins that drive tumorigenesis, such as downstream effectors of KRAS, the most commonly mutated human oncogene. As a consequence, several small-molecule HSP90 inhibitors are being evaluated in clinical trials as anticancer agents. To prospectively identify mechanisms through which HSP90-dependent cancer cells evade pharmacologic HSP90 blockade, we generated multiple mutant KRAS-driven cancer cell lines with acquired resistance to the purine-scaffold HSP90 inhibitor PU-H71. All cell lines retained dependence on HSP90 function, as evidenced by sensitivity to short hairpin RNA-mediated suppression of HSP90AA1 or HSP90AB1 (also called HSP90α and HSP90β, respectively), and exhibited two types of genomic alterations that interfere with the effects of PU-H71 on cell viability and proliferation: (i) a Y142N missense mutation in the ATP-binding domain of HSP90α that co-occurred with amplification of the HSP90AA1 locus, (ii) genomic amplification and overexpression of the ABCB1 gene encoding the MDR1 drug efflux pump. In support of a functional role for these alterations, exogenous expression of HSP90α Y142N conferred PU-H71 resistance to HSP90-dependent cells, and pharmacologic MDR1 inhibition with tariquidar or lowering ABCB1 expression restored sensitivity to PU-H71 in ABCB1-amplified cells. Finally, comparison with structurally distinct HSP90 inhibitors currently in clinical development revealed that PU-H71 resistance could be overcome, in part, by ganetespib (also known as STA9090) but not tanespimycin (also known as 17-AAG). Together, these data identify potential mechanisms of acquired resistance to small molecules targeting HSP90 that may warrant proactive screening for additional HSP90 inhibitors or rational combination therapies.