We developed a protocol to objectively quantify, via image analysis, the results of antigen retrieval (AR) maneuvers on immunohistochemical results for detection of P-glycoprotein (P-gp) in paraffin embedded test samples. Two carcinoma cell lines, KB 3-1 and KB V-1 (previously described non-expresser and high expresser of P-glycoprotein, respectively), were harvested, fixed for 4, 8, 12, 24 and 48 hr in formalin, placed in collodion bags, embedded into a single paraffin block, subjected to various antigen retrieval techniques, and stained with the anti-P-gp antibody C494. The amount of immunostaining was quantitated with the CAS 2.0 cell measurement program and optical density ratios were compared between the positive and negative cell lines. In overfixed cells (48 hr, 10% formalin), AR with microwaves and/or trypsin or other enzymes proved effective in optimizing P-gp detection as evidenced by an increased optical density ratio. In contrast, quantitation of staining of optimally fixed cells (4 hr in 10% formalin) demonstrated that AR decreased the optical density ratio. A postfixation step after the primary antibody also served to decrease the optical density ratio, via an increase in noise in the non-expressing cell. The use of optical density analysis by image computerization allowed objective, exquisitely sensitive, quantitative assessment of AR methods. (The J Histotechnol 22:9, 1999)