Cancer antigen-125 (CA-25) is the only marker acknowledged to have an established role in diagnosing epithelial ovarian cancer, monitoring its treatment, and detecting relapses. The optimal upper limit of the normal range for CA-125 in screening for ovarian cancer is, however, not clear. Some believe that a level exceeding 35 U/ml has predictive value in diagnosing ovarian cancer. The investigators attempted to determine the prevalence of ovarian cancer in 48,027 women aged 45-85 years who enrolled in the Shizuoka Cohort Study on Ovarian Cancer Screening trial. Of these women, 40,801 never had a CA-125 level greater than 35 U/ml and underwent transvaginal ultrasonography. CA-125 levels were measured by an enzyme-linked immunosorbent assay. Of the 40,801 women in the study, 4859 had abnormal pelvic ultrasound findings: 4741 in category 1 with simple morphology and 118 in category 2 with complex morphology. Surgery was carried out on 912 category 1 and 69 category 2 women. Ovarian cancer was diagnosed in 8 women (0.815%). Five of the 8 cancers (63%) were stage 1 lesions. The prevalence of ovarian cancer associated with abnormal ultrasound findings rose from 0.207% in women having a CA-125 level of up to 15 U/ml to 6.12% among those with CA-125 levels of 30-35 U/ml. Logistic regression analysis demonstrated that the CA-125 level significantly affected the risk of ovarian cancer. The odds ratio for ovarian cancer was 3.88 per unit increase in the CA-125 level (95% confidence interval, 3.49-4.37; P < 0.001). These findings suggest that surgically-detected ovarian cancer is not rare in women whose CA-125 levels are 35 U/ml or less-levels formerly considered to be within the normal range.