Most human ovarian carcinomas express mesothelin, which is shed as a diagnostically useful biomarker. We applied an ELISA to measure antibodies to native mesothelin in serum from a series of patients with divergent clinical outcomes. The level of anti-mesothelin antibodies determined as OD(450 nm) and referred to as absorption units (AU) for 1:20 diluted serum was higher in patients who remained disease-free after therapy [no evidence of disease (NED); n = 14] than in patients whose disease recurred [clinical evidence of disease (CED); n = 21; P < 0.01]. Applying AU > or = 0.5 at a serum dilution of 1:20 as cutoff, 10 of 14 (71%) ovarian carcinoma patients with NED and 9 of 21 (43%) patients with CED had antibodies to mesothelin compared with 6 of 23 (26%) healthy women (P < 0.008) and 5 of 24 (21%) women with other benign gynecologic diseases (P < 0.003), whereas 7 of 9 (78%) of women with pelvic inflammatory disease were positive. Three of the 14 (21%) NED patients had circulating mesothelin detected as an AU > or = 0.2 at a serum dilution of 1:40 (P < 0.005) compared with 15 of 21 (71%) CED patients, and 9 of 14 (64%) NED patients (P < 0.0002) were positive for antibodies and negative for antigen compared with 1 of 21 (5%) CED patients. Although our data indicate that an antibody response to mesothelin is an important correlate of ovarian carcinoma, prospective studies are needed to show whether the measurement of such antibodies (alone or together with antigen) aids the diagnosis and monitoring of patients.