To test the hypothesis that levels of adrenal autoantibodies correlate with the degree of adrenal dysfunction, we followed up adrenal cortex autoantibody (ACA) titers and 21-hydroxylase (21OH) autoantibody (21OHAb) levels in 19 ACA-positive subjects with preclinical Addison's disease. On enrollment, all the 19 ACA-positive subjects were positive for 21OHAb. At follow-up, the concordance rate for simultaneous presence/absence of both ACA and 21OHAb was as high as 91% and a strong, positive correlation between 21OHAb levels and ACA titers was observed (P < 0.0001). The levels of adrenal autoantibodies were positively associated with the severity of adrenal dysfunction (ANOVA, P < 0.0001 for both 21OHAb and ACA): the 21OH index was significantly lower at stage 0 or 1 than at stage 2+3 (corrected P < 0.001 andP < 0.05) or stage 4 (corrected P < 0.001 and <0.01). Similarly, ACA titer at stage 4 was significantly higher than stage 0 (P < 0.001), stage 1 (P < 0.001), and stage 2+3 (P < 0.05); and ACA titer at stage 2+3 was higher than stage 0 (P < 0.001) and stage 1 (P < 0.05). In subjects with progression of adrenal dysfunction (n = 14), levels of 21OHAb and ACA increased significantly (P = 0.041 and P = 0.002) during the follow-up period. In 5 subjects, the remission of biochemical signs of adrenal dysfunction was associated with the disappearance of both ACA and 21OHAb. Our study shows that the levels of adrenal autoantibodies correlate with the degree of adrenal dysfunction, and this suggests that production of high-level 21OHAb strongly signals the destructive phase of the autoimmune disease process.