Eight hundred and five consecutive patients with definite or classic rheumatoid arthritis, representative of a general population, were followed prospectively over an average of 12 years with 94 percent follow-up, during which time 233 patients died. Mortality rates were increased in rheumatoid arthritis by approximately 50 percent compared with population controls. "Excess deaths" (78) were due, in part, to disease-related causes, infections, and gastrointestinal problems secondary to therapy. Thirty-one deaths (14.1 percent) were due to malignancy, compared with 22.8 percent in non-age-adjusted population controls. Frequency of malignancy type (colon seven, lung four, breast three, prostate three, leukemia three, ovary three, stomach two, and lymphoma zero) did not differ significantly from the general population. Age-adjusted figures showed lower malignancy rates for patients with rheumatoid arthritis except after age 80. Two treatments were suitable for analysis. Gold-treated patients trended toward lower malignancy rates (11 versus 17 percent) than did those not treated with gold, and prednisone-treated patients had less malignancy (11 versus 20 percent) than did those not treated with prednisone. These data indicate that there is not an important increase in deaths due to malignancy in rheumatoid arthritis, and that at least two treatments, gold and prednisone, are not associated with higher risk of malignancy.