Bacterial vaccines have made dramatic impacts upon morbidity and mortality caused by a number of common pathogens, but a vaccine to prevent Staphylococcus aureus infections has proven to be illusive. With successful bacterial vaccines, the organisms are all part of the transient flora, whereas, S. aureus is part of the normal human flora. This means that S. aureus has had a prolonged time to adapt to the host milieu and its defences. The failure of several staphylococcal antigens to protect humans from infection in vaccine clinical trials using active or passive immunisation has stimulated a re-examination of the fundamental assumptions about staphylococcal immunity in humans vs. animals, especially rodents. This has spurred an active debate about the appropriate models for vaccine development and an examination of our current understanding of the protective immunity in humans. A major factor in the development of previous bacterial vaccines was a biomarker that predicted human protection, e.g., antibodies to tetanus toxoid or to pneumococcal polysaccharide. While antibodies against a number of staphylococcal antigens have proven to be an excellent biomarker for protection in rodents, these have not been translated to human infections. Thus, while much work remains, there is a growing consensus that T-cell immunity plays an important role in protecting humans. Moreover, the presence of anti-staphylococcal toxin antibodies correlates with reduced disease severity in humans. The most important recent advances concerning potential biomarkers, and the role of pre-existing immune status of vaccines in vaccine-associated mortality are considered in this review.