During the last decade, the outlook for patients with rheumatoid arthritis (RA) has improved greatly, especially for patients with autoantibody-positive RA. To further improve long-term disease outcomes, the field has turned to investigating the efficacy of treatment initiated in the pre-arthritic phase of RA, based on the adage “the sooner the better.” In this review, the concept of prevention is evaluated and different risk stages are being examined for their pre-test risks of RA development. These risks affect the post-test risk of biomarkers used at these stages and, consequently, the accuracy with which the risk of RA can be estimated. Furthermore, through their effect on accurate risk stratification, these pre-test risks ultimately also associate with the likelihood of false-negative trial results (the so-called “clinicostatistical tragedy”). Outcome measures to assess preventive effects are evaluated and relate to either the occurrence of disease itself or to the severity of risk factors for RA development. Results of recently completed prevention studies are discussed in the light of these theoretical considerations. The results vary, but clear prevention of RA has not been demonstrated. While some treatments (e.g. methotrexate) persistently reduced symptom severity, physical disability, and the severity of imaging joint inflammation, other treatments were not reported to produce long-lasting effects (hydroxychloroquine, rituximab, atorvastatin). The review concludes with future perspectives regarding the design of new prevention studies and considerations and requirements before findings can be implemented in daily practice in individuals at risk of RA attending rheumatology practices.