One of the biggest problems in immunohistochemistry has been how to maintain both good morphology and the immunoreactivity of antigens in tissue sections. Various techniques to retrieve the immunoreactivities of antigens (unmasking) after routine tissue preparations, such as fixation, dehydration and embedding, have been devised and are now finding their way into use for immunostainings in not only cytohistological investigations but also pathoclinical diagnoses. In this report, first, the mechanisms and significance of both fixation and antigen retrieval were surveyed from the viewpoint of protein inactivation. Second, some practical problems and notes in two of the most popular unmasking techniques, enzyme digestion and heat-induced epitope retrieval (HIER), were reviewed in order to adapt the techniques precisely in immunohistochemistry. The major artifact induced by fixation is the masking of tissue antigens due to cross-linking among the amino-acid residues of proteins. It is important to choose a proper fixing condition for each antigen considering its biochemical nature and resistance to the fixation. (Table 2), and to keep the fixing conditions to a minimum so that the immunoreactivity of the antigen can be readily retrieved by various unmasking techniques (Table 3, Fig. 1). Antigen retrieval per se is the process causing protein denaturation in tissues, just like many other protein inactivation processes (Table 1). Enzyme digestion may etch the masking parts of proteins around an antigen to expose its epitope. Although enzyme digestion is relatively simple and its treatment condition easy to control, the results are not necessarily dramatic and consistent depending on the types or lots of enzymes. Thus, one must find his/her own digestion manual to achieve the best staining result for each antigen (e.g., Table 4). For example, pepsin digestion gave the best results in the immunostaining of bromo-deoxyuridine (BrdU) and proliferating cell nuclear antigen (PCNA), whereas other enzymes had little effect (Table 5). Heating may also cleave the polypeptide backbone and disrupt the cross-links produced by fixation. The heating effect on antigen retrieval is temperature-dependent and seems to be proportional to the product of temperature and time. In the case of PCNA immunostaining on paraformaldehyde-fixed paraffin-embedded sections, heating at 90 degrees C for at least 3 min was required, however, as the heating condition became more severe, non-specific background stains also increased (Table 6), which is one of the most serious problems in antigen retrieval (Fig. 2a, c). One possible choice for avoiding such undesirable results is the combination of suboptimal heating (at 80 degrees C for 10-15 min) and pepsin digestion (Fig. 2b, d). An important theoretical consideration for employing such a dramatic method of denaturation is whether the masked antigen epitopes can be adequately exposed without giving rise to false-positive (or false-negative) results with previously trusted antibodies. It seems that each antigen requires a "tailor-made" tissue preparation for optimal preservation of its antigenicity and precise localization. In accordance with the development of immunology, molecular biology, genetics or embryology, the need for multiple immunostaining in combination with other technologies like in situ hybridization should increase in order to analyze the spatial and functional relationships among various molecules in situ. Antigen retrieval would then become a powerful strategy.