A cohort of 263 patients who completed Health Assessment Questionnaires (HAQ) each year from 1981 to 1989 is examined. Forty-three (43) possible covariates measured in 1981 are considered that might help predict mortality over the 8 years of the study. We use univariate correlation coefficients, stepwise regressions, and survival functions to identify a number of covariates that appear useful for predicting survival: age (-), prednisone use (-), disability index (-), male sex (-), never married (-), penicillamine use (+), divorced (+), and no occupation (-). Our results confirm studies suggesting that the HAQ disability index is a useful prognosticator of length of survival.