The molecular chaperone HSP90 has emerged as an exciting target for cancer treatment. We review the potential advantages of HSP90 inhibitors, particularly the simultaneous combinatorial depletion of multiple oncogenic "client" proteins, leading to blockade of many cancer-causing pathways and the antagonism of all of the hallmark pathological traits of malignancy. Cancer selectivity is achieved by exploiting cancer "dependencies," including oncogene addiction and the stressed state of malignant cells. The multiple downstream effects of HSP90 inhibitors should make the development of resistance more difficult than with agents having more restricted effects. We review the various classes of HSP90 inhibitor that have been developed, including the natural products geldanamycin and radicicol and also the purine scaffold and pyrazole/isoxazole class of synthetic small molecule inhibitors. A first-in-class HSP90 drug, the geldanamycin analog 17-AAG, has provided proof of concept for HSP90 inhibition in patients at well tolerated doses and therapeutic activity has been seen. Other inhibitors show promise in preclinical and clinical development. Opportunities and challenges for HSP90 inhibitors are discussed, including use in combination with other agents. Most of the current HSP90 inhibitors act by blocking the essential nucleotide binding and ATPase activity required for chaperone function. Potential new approaches are discussed, for example, interference with cochaperone binding and function in the superchaperone complex. Biomarkers for use with HSP90 inhibitors are described. We stress how basic and translational research has been mutually beneficial and indicate future directions to enhance our understanding of molecular chaperones and their exploitation in cancer and other diseases.