Treatment of rheumatoid arthritis (RA) presents a paradox: Many effective therapies are extensively documented in numerous clinical trials, but all long-term studies from clinical settings indicate that most patients with RA have progressive disease. This paradox may be explained in part on the basis of four observations: (1) Results of therapies for RA have been evaluated primarily according to clinical trials over short periods rather than according to long-term clinical observations over many years; (2) increased mortality rates in RA have not been widely recognized because RA is not included on death certificates of more than half of the patients who die with this disease; (3) mortality in RA generally has been attributed to causes unrelated to RA or to drug toxicity, but higher mortality in specific patients is predicted primarily by more severe clinical status; and (4) published reports indicating an optimistic prognosis in RA have been based on epidemiological studies rather than on patients in clinical settings. In epidemiological studies of large populations, 75% of individuals who meet the 1958 American Rheumatism Association (ARA) criteria for RA have no evidence of disease 3 to 5 years later; in contrast, in clinical settings more than 90% of patients with RA have evidence of disease 3 to 5 years later, generally with progression.