Osteoarthritis (OA) is the biggest unmet medical need among the many musculoskeletal conditions and the most common form of arthritis. It is a major cause of disability and impaired quality of life in the elderly. We review several ambitious but failed attempts to develop joint structure-modifying treatments for OA. Insights gleaned from these attempts suggest that these failures arose from unrealistic hypotheses, sub-optimal selection of patient populations or drug dose, and/or inadequate sensitivity of the trial endpoints. The long list of failures has prompted a paradigm shift in OA drug development with redirection of attention to: (1) consideration of the benefits of localized vs systemic pharmacological agents, as indicated by the increasing number of intra-articularly administered compounds entering clinical development; (2) recognition of OA as a complex disease with multiple phenotypes, that may each require somewhat different approaches for optimizing treatment; and (3) trial enhancements based on guidance regarding biomarkers provided by regulatory agencies, such as the Food and Drug Administration (FDA), that could be harnessed to help turn failures into successes.