We investigated whether circulating osteopontin (OPN) could be used as a biomarker for cervical cancer. We employed a monoclonal antibody (mAb 659) specific for the unique and intact thrombin-sensitive site in OPN using an inhibition ELISA. We found significantly higher levels of OPN in 33 cervical cancer patients in both the plasma (mean +/- SD, 612 +/- 106 ng/mL) and serum (424 +/- 121 ng/mL) compared to healthy subjects [409 +/- 56 ng/mL, from 31 plasma samples (P < 0.0001), and 314 +/- 98 ng/mL, from 32 serum samples (P = 0.0002), respectively]. Similar results were obtained when the plasma from a bigger group (147 individuals) of cervical cancer patients (560 +/- 211 ng/mL) were compared with the same plasma samples of the healthy individuals (P = 0.0014). More significantly, the OPN level was highest in stage III-IV disease (614 +/- 210 ng/mL, from 52 individuals; P = 0.0001) and least and non-discriminatory in stage I (473 +/- 110 ng/mL, from 40 individuals; P = 0.5318). No such discrimination was found when a mAb of a different specificity (mAb 446) was used in a similar inhibition ELISA to compare the two groups in the first study; a commercial capture ELISA also failed. The possibility that the target epitope recognized by the antibody probe in these assays was absent from the circulating OPN due to protein truncation was supported by gel fractionation of the OPN found in patients' plasma: 60-64 kDa fragments were found instead of the presumably full-length OPN (68 kDa) seen in healthy people. How these fragments are generated and what possible role they play in cancer biology remain interesting questions.