In this study, we quantitatively evaluated the intensity and extent of prostate-specific acid phosphatase in immunocytochemically stained prostate carcinoma tissue sections by using a dual-wave-length-based image processing technique. Tissue sections were doubly stained in a standard way, i.e., prostate-specific acid phosphatase was immunocytochemically labeled by peroxidase-antiperoxidase (PAP) technique using diaminobenezidene (DAB) as the substrate and hematoxylin as a nuclear counterstain. Statistical analysis in this study indicated that the quantitative measures of the prostate-specific acid phosphatase staining (PAP-DAB) are reproducible. We found that the quantitative intensity was generally proportional to the subjective intensity (graded as 1 to 4+) of PAP-DAB. Although the quantitative extent measurements compared with the subjective estimates of the percentage of tumor showing staining with PAP-DAB had a similar tendency, there were significant overlaps between extent of tumor staining falling in the mid-ranges. Because subjective grading of immunocytochemical prostate-specific acid phosphatase has been thought to be a useful marker of tumor differentiation in patients with prostate carcinoma, we also evaluated the relationship of the quantitative measures of PAP-DAB staining to other predictors of patient outcome, including histologic grades (Gleason score and MD Anderson grading scheme) and clinical stage. We confirmed that quantitative intensity and extent of PAP-DAB staining were independent of histologic grades and stage, just as the subjective measures of intensity and extent were found to be. Possible explanations of this lack of correlation are discussed.